id
string | submitter
string | authors
string | title
string | comments
string | journal-ref
string | doi
string | report-no
string | categories
string | license
string | abstract
string | versions
list | update_date
timestamp[s] | authors_parsed
sequence |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
2506.03971 | Moacir Aloisio | M. Aloisio, Silas L. Carvalho and C. R. de Oliveira | H\"older continuity and Fourier asymptotics of spectral measures for 1D Schr\"odinger operators under exponentially decaying perturbations | 41 pages. Minor revisions: typo corrections and textual refinements | null | null | null | math.SP math-ph math.CA math.MP | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | We establish $\frac{1}{2}$-H\"older continuity, or even the Lipschitz property, for the spectral measures of half-line discrete Schr\"odinger operators under suitable boundary conditions and exponentially decaying small potentials. These are the first known examples, apart from the free case, of Schr\"odinger operators with Lipschitz continuous spectral measures up to the spectral edge, and it was obtained as a consequence of the Dirichlet boundary condition. Notably, we show that the asymptotic behavior of the time-averaged quantum return probability, either $\log (t) / t$ or $1 / t$, as in the case of the free Laplacian, remains unchanged in this setting. Furthermore, we prove the persistence of the purely absolutely continuous spectrum and the $\frac{1}{2}$-H\"older continuity of the spectral measures for (Diophantine) quasi-periodic operators under exponentially decaying small perturbations. These results are optimal and hold for all energies, up to the border of the absolutely continuous spectrum. | [
{
"version": "v1",
"created": "Wed, 04 Jun 2025 14:01:54 GMT"
},
{
"version": "v2",
"created": "Wed, 11 Jun 2025 03:21:03 GMT"
}
] | 2025-06-12T00:00:00 | [
[
"Aloisio",
"M.",
""
],
[
"Carvalho",
"Silas L.",
""
],
[
"de Oliveira",
"C. R.",
""
]
] |
2506.03998 | Tara Liebisch | Klaus Blaum (Max Planck Institute for Nuclear Physics), Peter Kaufmann (German National Research and Education Network, DFN), Jochen Kronj\"ager (Physikalisch-Technische Bundesanstalt), Stefan K\"uck (Physikalisch-Technische Bundesanstalt), Tara Cubel Liebisch (Physikalisch-Technische Bundesanstalt), Dieter Meschede (University of Bonn), Susanne Naegele-Jackson (Friedrich-Alexander-Universit\"at Erlangen-N\"urnberg), Stephan Schiller (Heinrich Heine University D\"usseldorf), Harald Schnatz (Physikalisch-Technische Bundesanstalt) | The QTF-Backbone: Proposal for a Nationwide Optical Fibre Backbone in Germany for Quantum Technology and Time and Frequency Metrology | 52 pages, 7 figures, 9 tables, 73 contributors and 28 supporters in addition to the 9 authors | null | null | null | physics.ins-det physics.atom-ph physics.geo-ph quant-ph | http://creativecommons.org/licenses/by/4.0/ | The recent breakthroughs in the distribution of quantum information and high-precision time and frequency (T&F) signals over long-haul optical fibre networks have transformative potential for physically secure communications, resilience of Global Navigation Satellite Systems (GNSS) and fundamental physics. However, so far these capabilities remain confined to isolated testbeds, with quantum and T&F signals accessible, for example in Germany, to only a few institutions.
We propose the QTF-Backbone: a dedicated national fibre-optic infrastructure in Germany for the networked distribution of quantum and T&F signals using dark fibres and specialized hardware. The QTF-Backbone is planned as a four-phase deployment over ten years to ensure scalable, sustainable access for research institutions and industry. The concept builds on successful demonstrations of high-TRL time and frequency distribution across Europe, including PTB-MPQ links in Germany, REFIMEVE in France, and the Italian LIFT network. The QTF-Backbone will enable transformative R&D, support a nationwide QTF ecosystem, and ensure the transition from innovation to deployment. As a national and European hub, it will position Germany and Europe at the forefront of quantum networking, as well as time and frequency transfer. | [
{
"version": "v1",
"created": "Wed, 04 Jun 2025 14:27:50 GMT"
},
{
"version": "v2",
"created": "Wed, 11 Jun 2025 13:01:04 GMT"
}
] | 2025-06-12T00:00:00 | [
[
"Blaum",
"Klaus",
"",
"Max Planck Institute for Nuclear Physics"
],
[
"Kaufmann",
"Peter",
"",
"German National Research and Education Network, DFN"
],
[
"Kronjäger",
"Jochen",
"",
"Physikalisch-Technische Bundesanstalt"
],
[
"Kück",
"Stefan",
"",
"Physikalisch-Technische Bundesanstalt"
],
[
"Liebisch",
"Tara Cubel",
"",
"Physikalisch-Technische Bundesanstalt"
],
[
"Meschede",
"Dieter",
"",
"University of Bonn"
],
[
"Naegele-Jackson",
"Susanne",
"",
"Friedrich-Alexander-Universität Erlangen-Nürnberg"
],
[
"Schiller",
"Stephan",
"",
"Heinrich Heine University Düsseldorf"
],
[
"Schnatz",
"Harald",
"",
"Physikalisch-Technische Bundesanstalt"
]
] |
2506.04112 | Weitao Liu | Weitao Liu | The optical Bloch equation for the finite-temperature fluctuations | 10 pages, 4 figures | null | null | null | quant-ph | http://creativecommons.org/licenses/by/4.0/ | In this work, I analyze the quantum fluctuations and the thermal fluctuations in the framework of quantum mechanics. Being recognized as incoherent perturbations with different features, fluctuations of these two types lead to dissipative terms in the optical Bloch equations. The method allows one to use the optical Bloch equation to analyze time-dependent processes in the finite-temperature fluctuations. The numerical results show that the deexcitation is the limit of the equilibration at zero temperature. The impact of the fluctuations on the coherent excitations are also discussed. | [
{
"version": "v1",
"created": "Wed, 04 Jun 2025 16:05:06 GMT"
},
{
"version": "v2",
"created": "Fri, 06 Jun 2025 12:26:12 GMT"
},
{
"version": "v3",
"created": "Wed, 11 Jun 2025 07:48:50 GMT"
}
] | 2025-06-12T00:00:00 | [
[
"Liu",
"Weitao",
""
]
] |
2506.04278 | Yumiko Nishiyama | Yumiko Nishiyama | Redefining Functionality and Construction-Defining Capacity: Functions as Principles of Syntactic and Semantic Generation | This is a revised version for submission to LMCS | null | null | null | cs.LO | http://creativecommons.org/licenses/by/4.0/ | This study redefines the notion of functionality-traditionally understood as a property of mappings or structure preservation-from a more fundamental and generative perspective. Introducing the concept of a Construction-Defining function (CDF), we formalize functionality as a dual capacity to generate both syntactic terms and semantic interpretations. We provide an explicit axiomatization of CDF based on syntactic generativity and semantic compositionality, and further construct categorical models using initial algebras and endofunctors to validate the generality of this concept. Through comparative analysis with type theory, model theory, and formal semantics, we demonstrate that functionality, in this enriched sense, serves as a foundational principle for structural generation across diverse theoretical domains. By reexamining the very nature of functions and their role in structure formation, this work contributes to a unified understanding of logic,semantics, and computation. | [
{
"version": "v1",
"created": "Wed, 04 Jun 2025 02:34:22 GMT"
},
{
"version": "v2",
"created": "Tue, 10 Jun 2025 21:45:47 GMT"
}
] | 2025-06-12T00:00:00 | [
[
"Nishiyama",
"Yumiko",
""
]
] |
2506.04394 | Qiuyu Tang | Qiuyu Tang, Bonor Ayambem, Mooi Choo Chuah, Aparna Bharati | Is Perturbation-Based Image Protection Disruptive to Image Editing? | 6 pages, 8 figures, accepted by ICIP 2025 | null | null | null | cs.CV | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | The remarkable image generation capabilities of state-of-the-art diffusion models, such as Stable Diffusion, can also be misused to spread misinformation and plagiarize copyrighted materials. To mitigate the potential risks associated with image editing, current image protection methods rely on adding imperceptible perturbations to images to obstruct diffusion-based editing. A fully successful protection for an image implies that the output of editing attempts is an undesirable, noisy image which is completely unrelated to the reference image. In our experiments with various perturbation-based image protection methods across multiple domains (natural scene images and artworks) and editing tasks (image-to-image generation and style editing), we discover that such protection does not achieve this goal completely. In most scenarios, diffusion-based editing of protected images generates a desirable output image which adheres precisely to the guidance prompt. Our findings suggest that adding noise to images may paradoxically increase their association with given text prompts during the generation process, leading to unintended consequences such as better resultant edits. Hence, we argue that perturbation-based methods may not provide a sufficient solution for robust image protection against diffusion-based editing. | [
{
"version": "v1",
"created": "Wed, 04 Jun 2025 19:20:37 GMT"
},
{
"version": "v2",
"created": "Tue, 10 Jun 2025 19:45:40 GMT"
}
] | 2025-06-12T00:00:00 | [
[
"Tang",
"Qiuyu",
""
],
[
"Ayambem",
"Bonor",
""
],
[
"Chuah",
"Mooi Choo",
""
],
[
"Bharati",
"Aparna",
""
]
] |
2506.04420 | Kareem Elgindy | Kareem T. Elgindy | Sustainable Water Treatment through Fractional-Order Chemostat Modeling with Sliding Memory and Periodic Boundary Conditions: A Mathematical Framework for Clean Water and Sanitation | null | null | null | null | math.OC math.DS | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | This study investigates the theoretical properties of a fractional-order chemostat system with sliding memory and periodic boundary conditions, used to model the cultivation of microorganisms for pollutant degradation. By incorporating Caputo fractional derivatives with a sliding memory window (CFDS), the model captures time-dependent behaviors and memory effects in biological systems more effectively than integer-order derivatives. We reduce the two-dimensional fractional differential equations (FDEs) governing substrate and biomass concentrations to a one-dimensional FDE, utilizing periodic boundary conditions. The existence and uniqueness of non-constant periodic solutions are established using the Carath\'eodory framework and fixed-point theorems, ensuring the system's well-posedness. We prove the positivity and boundedness of solutions, demonstrating that substrate concentrations remain within physically meaningful bounds and biomass concentrations stay strictly positive, with solution trajectories confined to a biologically feasible invariant set. Additionally, we analyze non-trivial equilibria under constant dilution rates and derive their stability properties. The results validate the mathematical robustness of the fractional-order chemostat model for periodic bioprocesses, offering a foundation for advanced water treatment technologies and sanitation improvements. | [
{
"version": "v1",
"created": "Wed, 04 Jun 2025 20:02:57 GMT"
},
{
"version": "v2",
"created": "Wed, 11 Jun 2025 11:49:13 GMT"
}
] | 2025-06-12T00:00:00 | [
[
"Elgindy",
"Kareem T.",
""
]
] |
2506.04430 | Andrey Veprikov | Egor Petrov, Grigoriy Evseev, Aleksey Antonov, Andrey Veprikov, Pavel Plyusnin, Nikolay Bushkov, Stanislav Moiseev and Aleksandr Beznosikov | Leveraging Coordinate Momentum in SignSGD and Muon: Memory-Optimized Zero-Order | 26 pages, 5 tables | null | null | null | cs.LG math.OC | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Fine-tuning Large Language Models (LLMs) is essential for adapting pre-trained models to downstream tasks. Yet traditional first-order optimizers such as Stochastic Gradient Descent (SGD) and Adam incur prohibitive memory and computational costs that scale poorly with model size. In this paper, we investigate zero-order (ZO) optimization methods as a memory- and compute-efficient alternative, particularly in the context of parameter-efficient fine-tuning techniques like LoRA. We propose $\texttt{JAGUAR SignSGD}$, a ZO momentum-based algorithm that extends ZO SignSGD, requiring the same number of parameters as the standard ZO SGD and only $\mathcal{O}(1)$ function evaluations per iteration. To the best of our knowledge, this is the first study to establish rigorous convergence guarantees for SignSGD in the stochastic ZO case. We further propose $\texttt{JAGUAR Muon}$, a novel ZO extension of the Muon optimizer that leverages the matrix structure of model parameters, and we provide its convergence rate under arbitrary stochastic noise. Through extensive experiments on challenging LLM fine-tuning benchmarks, we demonstrate that the proposed algorithms meet or exceed the convergence quality of standard first-order methods, achieving significant memory reduction. Our theoretical and empirical results establish new ZO optimization methods as a practical and theoretically grounded approach for resource-constrained LLM adaptation. Our code is available at https://github.com/brain-mmo-lab/ZO_LLM | [
{
"version": "v1",
"created": "Wed, 04 Jun 2025 20:27:17 GMT"
},
{
"version": "v2",
"created": "Wed, 11 Jun 2025 17:05:40 GMT"
}
] | 2025-06-12T00:00:00 | [
[
"Petrov",
"Egor",
""
],
[
"Evseev",
"Grigoriy",
""
],
[
"Antonov",
"Aleksey",
""
],
[
"Veprikov",
"Andrey",
""
],
[
"Plyusnin",
"Pavel",
""
],
[
"Bushkov",
"Nikolay",
""
],
[
"Moiseev",
"Stanislav",
""
],
[
"Beznosikov",
"Aleksandr",
""
]
] |
2506.04593 | Qiong Wu | Xun Li, Qiong Wu, Pingyi Fan, Kezhi Wang, Nan Cheng, Khaled B. Letaief | Federated Learning Assisted Edge Caching Scheme Based on Lightweight Architecture DDPM | This paper has been submitted to IEEE letters. The source code has been released at: https://github.com/qiongwu86/Federated-Learning-Assisted-Edge-Caching-Scheme-Based-on-Lightweight-Architecture-DDPM | null | null | null | cs.NI eess.SP | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Edge caching is an emerging technology that empowers caching units at edge nodes, allowing users to fetch contents of interest that have been pre-cached at the edge nodes. The key to pre-caching is to maximize the cache hit percentage for cached content without compromising users' privacy. In this letter, we propose a federated learning (FL) assisted edge caching scheme based on lightweight architecture denoising diffusion probabilistic model (LDPM). Our simulation results verify that our proposed scheme achieves a higher cache hit percentage compared to existing FL-based methods and baseline methods. | [
{
"version": "v1",
"created": "Thu, 05 Jun 2025 03:16:51 GMT"
},
{
"version": "v2",
"created": "Wed, 11 Jun 2025 02:24:07 GMT"
}
] | 2025-06-12T00:00:00 | [
[
"Li",
"Xun",
""
],
[
"Wu",
"Qiong",
""
],
[
"Fan",
"Pingyi",
""
],
[
"Wang",
"Kezhi",
""
],
[
"Cheng",
"Nan",
""
],
[
"Letaief",
"Khaled B.",
""
]
] |
2506.04704 | Youngwan Lee | Youngwan Lee, Kangsan Kim, Kwanyong Park, Ilcahe Jung, Soojin Jang, Seanie Lee, Yong-Ju Lee, Sung Ju Hwang | HoliSafe: Holistic Safety Benchmarking and Modeling with Safety Meta Token for Vision-Language Model | Project page: https://youngwanlee.github.io/holisafe | null | null | null | cs.CV cs.AI | http://creativecommons.org/licenses/by-nc-nd/4.0/ | Despite emerging efforts to enhance the safety of Vision-Language Models (VLMs), current approaches face two main shortcomings. 1) Existing safety-tuning datasets and benchmarks only partially consider how image-text interactions can yield harmful content, often overlooking contextually unsafe outcomes from seemingly benign pairs. This narrow coverage leaves VLMs vulnerable to jailbreak attacks in unseen configurations. 2) Prior methods rely primarily on data-centric tuning, with limited architectural innovations to intrinsically strengthen safety. We address these gaps by introducing a holistic safety dataset and benchmark, HoliSafe, that spans all five safe/unsafe image-text combinations, providing a more robust basis for both training and evaluation. We further propose SafeLLaVA, a novel VLM augmented with a learnable safety meta token and a dedicated safety head. The meta token encodes harmful visual cues during training, intrinsically guiding the language model toward safer responses, while the safety head offers interpretable harmfulness classification aligned with refusal rationales. Experiments show that SafeLLaVA, trained on HoliSafe, achieves state-of-the-art safety performance across multiple VLM benchmarks. Additionally, the HoliSafe benchmark itself reveals critical vulnerabilities in existing models. We hope that HoliSafe and SafeLLaVA will spur further research into robust and interpretable VLM safety, expanding future avenues for multimodal alignment. | [
{
"version": "v1",
"created": "Thu, 05 Jun 2025 07:26:34 GMT"
},
{
"version": "v2",
"created": "Wed, 11 Jun 2025 12:53:30 GMT"
}
] | 2025-06-12T00:00:00 | [
[
"Lee",
"Youngwan",
""
],
[
"Kim",
"Kangsan",
""
],
[
"Park",
"Kwanyong",
""
],
[
"Jung",
"Ilcahe",
""
],
[
"Jang",
"Soojin",
""
],
[
"Lee",
"Seanie",
""
],
[
"Lee",
"Yong-Ju",
""
],
[
"Hwang",
"Sung Ju",
""
]
] |
2506.04900 | Yuhao Li | Cui Rui, Li Yuhao, Song Xiaojun | Power-boosting in Specification Tests using Kernel Directional Component | null | null | null | null | econ.EM | http://creativecommons.org/licenses/by/4.0/ | We propose power-boosting strategies for kernel-based specification tests in conditional moment models, with a focus on the Kernel Conditional Moment (KCM) test. By decomposing the KCM statistic into spectral components, we demonstrate that truncating poorly estimated directions and selecting kernels based on a non-asymptotic signal-to-noise ratio significantly improves both test power and size control. Our theoretical and simulation results demonstrate that, while divergent component weights may offer higher asymptotic power, convergent component weights perform better in finite samples. The methods outperform existing tests across various settings and are illustrated in an empirical application. | [
{
"version": "v1",
"created": "Thu, 05 Jun 2025 11:30:57 GMT"
}
] | 2025-06-12T00:00:00 | [
[
"Rui",
"Cui",
""
],
[
"Yuhao",
"Li",
""
],
[
"Xiaojun",
"Song",
""
]
] |
2506.05176 | Mingxin Li | Yanzhao Zhang, Mingxin Li, Dingkun Long, Xin Zhang, Huan Lin, Baosong Yang, Pengjun Xie, An Yang, Dayiheng Liu, Junyang Lin, Fei Huang, Jingren Zhou | Qwen3 Embedding: Advancing Text Embedding and Reranking Through Foundation Models | null | null | null | null | cs.CL | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | In this work, we introduce the Qwen3 Embedding series, a significant advancement over its predecessor, the GTE-Qwen series, in text embedding and reranking capabilities, built upon the Qwen3 foundation models. Leveraging the Qwen3 LLMs' robust capabilities in multilingual text understanding and generation, our innovative multi-stage training pipeline combines large-scale unsupervised pre-training with supervised fine-tuning on high-quality datasets. Effective model merging strategies further ensure the robustness and adaptability of the Qwen3 Embedding series. During the training process, the Qwen3 LLMs serve not only as backbone models but also play a crucial role in synthesizing high-quality, rich, and diverse training data across multiple domains and languages, thus enhancing the training pipeline. The Qwen3 Embedding series offers a spectrum of model sizes (0.6B, 4B, 8B) for both embedding and reranking tasks, addressing diverse deployment scenarios where users can optimize for either efficiency or effectiveness. Empirical evaluations demonstrate that the Qwen3 Embedding series achieves state-of-the-art results across diverse benchmarks. Notably, it excels on the multilingual evaluation benchmark MTEB for text embedding, as well as in various retrieval tasks, including code retrieval, cross-lingual retrieval and multilingual retrieval. To facilitate reproducibility and promote community-driven research and development, the Qwen3 Embedding models are publicly available under the Apache 2.0 license. | [
{
"version": "v1",
"created": "Thu, 05 Jun 2025 15:49:48 GMT"
},
{
"version": "v2",
"created": "Tue, 10 Jun 2025 06:03:06 GMT"
},
{
"version": "v3",
"created": "Wed, 11 Jun 2025 02:54:49 GMT"
}
] | 2025-06-12T00:00:00 | [
[
"Zhang",
"Yanzhao",
""
],
[
"Li",
"Mingxin",
""
],
[
"Long",
"Dingkun",
""
],
[
"Zhang",
"Xin",
""
],
[
"Lin",
"Huan",
""
],
[
"Yang",
"Baosong",
""
],
[
"Xie",
"Pengjun",
""
],
[
"Yang",
"An",
""
],
[
"Liu",
"Dayiheng",
""
],
[
"Lin",
"Junyang",
""
],
[
"Huang",
"Fei",
""
],
[
"Zhou",
"Jingren",
""
]
] |
2506.05224 | Jose L. Goity | D. Jayakodige and J. L. Goity | Pion-Nucleon Scattering in Baryon Chiral Perturbation Theory combined with the ${ 1/N_c}$ Expansion | 54 pages, 11 figures | null | null | JLAB-THY-25-4343 | hep-ph nucl-th | http://creativecommons.org/licenses/by-nc-nd/4.0/ | This work implements the combined BChPT and 1/Nc expansions for pion-nucleon elastic scattering. The effective theory is based on the baryon sector dynamical spin-flavor SU(4) symmetry emergent in the large Nc limit, whose breaking is controlled by the $1/N_c$ expansion. The non-commutativity of the chiral and 1/Nc expansions in unitarity corrections (loops) requires a linking of both expansions. As it was shown in the case of baryon masses and currents, the natural linking is the $\xi$-expansion, in which $O(p) = O(1/Nc ) = O(\xi)A$. The spin-flavor symmetry requires that the ground state baryons span an SU(4) symmetric irreducible representation which implies that in particular $N$ and $\Delta$ are active degrees of freedom in the effective theory. The scattering amplitude is expanded to the next-to-next-to leading order in the $\xi$ expansion, corresponding to the one-loop contributions with the LO Lagrangian. The results are given for generic $N_c$ in order to demonstrate the consistency of the framework. The spin-flavor symmetry plays a central role in maintaining the consistency of the effective theory with respect to the $1/N_c$ expansion. This consistency manifests itself in an improvement in the convergence of the low energy expansion with respect to the case of the ordinary BChPT without an explicit dynamical $\Delta$, which is known to be inconsistent with the constraints of $N_c$ scaling. Fits to the $\pi N \to \pi N$ S, P and D partial wave amplitudes from the SAID data base are finally used to test the framework and to determine the energy range of its applicability. | [
{
"version": "v1",
"created": "Thu, 05 Jun 2025 16:39:39 GMT"
},
{
"version": "v2",
"created": "Wed, 11 Jun 2025 17:34:04 GMT"
}
] | 2025-06-12T00:00:00 | [
[
"Jayakodige",
"D.",
""
],
[
"Goity",
"J. L.",
""
]
] |
2506.05343 | Renjie Chen | Wenfeng Lin, Renjie Chen, Boyuan Liu, Shiyue Yan, Ruoyu Feng, Jiangchuan Wei, Yichen Zhang, Yimeng Zhou, Chao Feng, Jiao Ran, Qi Wu, Zuotao Liu, Mingyu Guo | ContentV: Efficient Training of Video Generation Models with Limited Compute | Project Page: https://contentv.github.io | null | null | null | cs.CV | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Recent advances in video generation demand increasingly efficient training recipes to mitigate escalating computational costs. In this report, we present ContentV, an 8B-parameter text-to-video model that achieves state-of-the-art performance (85.14 on VBench) after training on 256 x 64GB Neural Processing Units (NPUs) for merely four weeks. ContentV generates diverse, high-quality videos across multiple resolutions and durations from text prompts, enabled by three key innovations: (1) A minimalist architecture that maximizes reuse of pre-trained image generation models for video generation; (2) A systematic multi-stage training strategy leveraging flow matching for enhanced efficiency; and (3) A cost-effective reinforcement learning with human feedback framework that improves generation quality without requiring additional human annotations. All the code and models are available at: https://contentv.github.io. | [
{
"version": "v1",
"created": "Thu, 05 Jun 2025 17:59:54 GMT"
},
{
"version": "v2",
"created": "Wed, 11 Jun 2025 15:48:38 GMT"
}
] | 2025-06-12T00:00:00 | [
[
"Lin",
"Wenfeng",
""
],
[
"Chen",
"Renjie",
""
],
[
"Liu",
"Boyuan",
""
],
[
"Yan",
"Shiyue",
""
],
[
"Feng",
"Ruoyu",
""
],
[
"Wei",
"Jiangchuan",
""
],
[
"Zhang",
"Yichen",
""
],
[
"Zhou",
"Yimeng",
""
],
[
"Feng",
"Chao",
""
],
[
"Ran",
"Jiao",
""
],
[
"Wu",
"Qi",
""
],
[
"Liu",
"Zuotao",
""
],
[
"Guo",
"Mingyu",
""
]
] |
2506.05387 | Jaydip Sen Prof. | Jaydip Sen, Saptarshi Sengupta, Subhasis Dasgupta | Advancing Decoding Strategies: Enhancements in Locally Typical Sampling for LLMs | This is the accepted but pre-reviewed version of the chapter that has been accepted for publication in the Springer volume 'Decision-Making in Computational Intelligence-Based Systems,' edited by Witold Pedrycz, Gilberto Rivera, Rose Ma Rodriguez, and Salvador Ibarra Martinez. The chapter is 39 pages long, and it contains 2 figures and 6 tables. This is NOT the final camera-ready version | null | null | null | cs.CL cs.AI | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | This chapter explores advancements in decoding strategies for large language models (LLMs), focusing on enhancing the Locally Typical Sampling (LTS) algorithm. Traditional decoding methods, such as top-k and nucleus sampling, often struggle to balance fluency, diversity, and coherence in text generation. To address these challenges, Adaptive Semantic-Aware Typicality Sampling (ASTS) is proposed as an improved version of LTS, incorporating dynamic entropy thresholding, multi-objective scoring, and reward-penalty adjustments. ASTS ensures contextually coherent and diverse text generation while maintaining computational efficiency. Its performance is evaluated across multiple benchmarks, including story generation and abstractive summarization, using metrics such as perplexity, MAUVE, and diversity scores. Experimental results demonstrate that ASTS outperforms existing sampling techniques by reducing repetition, enhancing semantic alignment, and improving fluency. | [
{
"version": "v1",
"created": "Tue, 03 Jun 2025 14:25:23 GMT"
},
{
"version": "v2",
"created": "Wed, 11 Jun 2025 16:08:29 GMT"
}
] | 2025-06-12T00:00:00 | [
[
"Sen",
"Jaydip",
""
],
[
"Sengupta",
"Saptarshi",
""
],
[
"Dasgupta",
"Subhasis",
""
]
] |
2506.05408 | Jonathan Scott | Jonathan Scott, Christoph H. Lampert, David Saulpic | Differentially Private Federated $k$-Means Clustering with Server-Side Data | null | null | null | null | cs.CR cs.LG | http://creativecommons.org/licenses/by/4.0/ | Clustering is a cornerstone of data analysis that is particularly suited to identifying coherent subgroups or substructures in unlabeled data, as are generated continuously in large amounts these days. However, in many cases traditional clustering methods are not applicable, because data are increasingly being produced and stored in a distributed way, e.g. on edge devices, and privacy concerns prevent it from being transferred to a central server. To address this challenge, we present FedDP-KMeans, a new algorithm for $k$-means clustering that is fully-federated as well as differentially private. Our approach leverages (potentially small and out-of-distribution) server-side data to overcome the primary challenge of differentially private clustering methods: the need for a good initialization. Combining our initialization with a simple federated DP-Lloyds algorithm we obtain an algorithm that achieves excellent results on synthetic and real-world benchmark tasks. We also provide a theoretical analysis of our method that provides bounds on the convergence speed and cluster identification success. | [
{
"version": "v1",
"created": "Wed, 04 Jun 2025 14:53:25 GMT"
},
{
"version": "v2",
"created": "Wed, 11 Jun 2025 09:31:04 GMT"
}
] | 2025-06-12T00:00:00 | [
[
"Scott",
"Jonathan",
""
],
[
"Lampert",
"Christoph H.",
""
],
[
"Saulpic",
"David",
""
]
] |
2506.05504 | Pei-Hao Fu | Pei-Hao Fu, Qianqian Lv, Yong Xu, Jorge Cayao, Jun-Feng Liu, and Xiang-Long Yu | All-electrically controlled spintronics in altermagnetic heterostructures | 13 pages, 5 figures | null | null | null | cond-mat.mes-hall cond-mat.mtrl-sci physics.app-ph | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | The recent development of altermagnetic materials, supporting spin splitting without net magnetization, opens new directions for spintronics that are fundamentally distinct from conventional ferromagnetic, antiferromagnetic, or spin-orbit coupling systems. Here we investigate spin-selective quantum transport in heterostructures composed of a normal metal and a two-dimensional $d$-wave altermagnet. We focus on two types of $d$-wave altermagnets, namely, weak and strong altermagnets that support close elliptic and open hyperbolic spin-resolved Fermi surfaces, respectively. Building on these distinct electronic structures, we propose all-electrically controlled spin filter and spin valve devices, where quantum resonant tunneling enables highly spin-polarized conductance tunable via gate voltage and interface transparency. In particular, we find that strong altermagnets allow gate-tunable full spin polarization that is robust against interface scattering and can be reversed by gate control. We further demonstrate that a double-gated spin valve electrically switches between parallel and antiparallel spin configurations, analogous to magnetic junctions but without the need for external magnetic fields. Our results establish both weak and strong altermagnets as promising platforms for realizing magnetic-field-free electrically tunable spintronic functionalities. | [
{
"version": "v1",
"created": "Thu, 05 Jun 2025 18:43:34 GMT"
},
{
"version": "v2",
"created": "Wed, 11 Jun 2025 13:59:52 GMT"
}
] | 2025-06-12T00:00:00 | [
[
"Fu",
"Pei-Hao",
""
],
[
"Lv",
"Qianqian",
""
],
[
"Xu",
"Yong",
""
],
[
"Cayao",
"Jorge",
""
],
[
"Liu",
"Jun-Feng",
""
],
[
"Yu",
"Xiang-Long",
""
]
] |
2506.05668 | Jiajun He | Jiajun He, Jos\'e Miguel Hern\'andez-Lobato, Yuanqi Du, Francisco Vargas | RNE: a plug-and-play framework for diffusion density estimation and inference-time control | 39 pages; 14 figures | null | null | null | cs.LG stat.ML | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | In this paper, we introduce the Radon-Nikodym Estimator (RNE), a flexible, plug-and-play framework for diffusion inference-time density estimation and control, based on the concept of the density ratio between path distributions. RNE connects and unifies a variety of existing density estimation and inference-time control methods under a single and intuitive perspective, stemming from basic variational inference and probabilistic principles therefore offering both theoretical clarity and practical versatility. Experiments demonstrate that RNE delivers strong results in diffusion density estimation, and offers broad applicability to inference-time control tasks -- such as annealing, diffusion model composition, and reward-tilting -- with promising inference-time scaling performance. | [
{
"version": "v1",
"created": "Fri, 06 Jun 2025 01:32:14 GMT"
},
{
"version": "v2",
"created": "Wed, 11 Jun 2025 16:51:44 GMT"
}
] | 2025-06-12T00:00:00 | [
[
"He",
"Jiajun",
""
],
[
"Hernández-Lobato",
"José Miguel",
""
],
[
"Du",
"Yuanqi",
""
],
[
"Vargas",
"Francisco",
""
]
] |
2506.05765 | Taiga Shinozaki | Taiga Shinozaki, Tomoki Doi, Amane Watahiki, Satoshi Nishida, Hitomi Yanaka | Do Large Vision-Language Models Distinguish between the Actual and Apparent Features of Illusions? | To appear in the Proceedings of the 47th Annual Meeting of the Cognitive Science Society (COGSCI 2025) | null | null | null | cs.CV cs.CL | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Humans are susceptible to optical illusions, which serve as valuable tools for investigating sensory and cognitive processes. Inspired by human vision studies, research has begun exploring whether machines, such as large vision language models (LVLMs), exhibit similar susceptibilities to visual illusions. However, studies often have used non-abstract images and have not distinguished actual and apparent features, leading to ambiguous assessments of machine cognition. To address these limitations, we introduce a visual question answering (VQA) dataset, categorized into genuine and fake illusions, along with corresponding control images. Genuine illusions present discrepancies between actual and apparent features, whereas fake illusions have the same actual and apparent features even though they look illusory due to the similar geometric configuration. We evaluate the performance of LVLMs for genuine and fake illusion VQA tasks and investigate whether the models discern actual and apparent features. Our findings indicate that although LVLMs may appear to recognize illusions by correctly answering questions about both feature types, they predict the same answers for both Genuine Illusion and Fake Illusion VQA questions. This suggests that their responses might be based on prior knowledge of illusions rather than genuine visual understanding. The dataset is available at https://github.com/ynklab/FILM | [
{
"version": "v1",
"created": "Fri, 06 Jun 2025 05:47:50 GMT"
},
{
"version": "v2",
"created": "Wed, 11 Jun 2025 02:03:57 GMT"
}
] | 2025-06-12T00:00:00 | [
[
"Shinozaki",
"Taiga",
""
],
[
"Doi",
"Tomoki",
""
],
[
"Watahiki",
"Amane",
""
],
[
"Nishida",
"Satoshi",
""
],
[
"Yanaka",
"Hitomi",
""
]
] |
2506.05794 | Luca M. Possati Dr. | Luca M. Possati | Markov Blanket Density and Free Energy Minimization | null | null | null | null | q-bio.NC cs.IT math.IT | http://creativecommons.org/licenses/by-nc-nd/4.0/ | This paper presents a continuous, information-theoretic extension of the Free Energy Principle through the concept of Markov blanket density, i.e., a scalar field that quantifies the degree of conditional independence between internal and external states at each point in space (ranging from 0 for full coupling to 1 for full separation). It demonstrates that active inference dynamics (including the minimization of variational and expected free energy) naturally emerge from spatial gradients in this density, making Markov blanket density a necessary foundation for the definability and coherence of the Free Energy Principle. These ideas are developed through a mathematically framework that links density gradients to precise and testable dynamics, offering a foundation for novel predictions and simulation paradigms. | [
{
"version": "v1",
"created": "Fri, 06 Jun 2025 06:39:46 GMT"
},
{
"version": "v2",
"created": "Wed, 11 Jun 2025 12:31:22 GMT"
}
] | 2025-06-12T00:00:00 | [
[
"Possati",
"Luca M.",
""
]
] |
2506.05802 | Adriana Stan PhD | Adriana Stan, David Combei, Dan Oneata, and Horia Cucu | TADA: Training-free Attribution and Out-of-Domain Detection of Audio Deepfakes | Accepted at Interspeech 2025 | null | null | null | eess.AS | http://creativecommons.org/licenses/by-nc-nd/4.0/ | Deepfake detection has gained significant attention across audio, text, and image modalities, with high accuracy in distinguishing real from fake. However, identifying the exact source--such as the system or model behind a deepfake--remains a less studied problem. In this paper, we take a significant step forward in audio deepfake model attribution or source tracing by proposing a training-free, green AI approach based entirely on k-Nearest Neighbors (kNN). Leveraging a pre-trained self-supervised learning (SSL) model, we show that grouping samples from the same generator is straightforward--we obtain an 0.93 F1-score across five deepfake datasets. The method also demonstrates strong out-of-domain (OOD) detection, effectively identifying samples from unseen models at an F1-score of 0.84.
We further analyse these results in a multi-dimensional approach and provide additional insights. All code and data protocols used in this work are available in our open repository: https://github.com/adrianastan/tada/. | [
{
"version": "v1",
"created": "Fri, 06 Jun 2025 07:00:23 GMT"
},
{
"version": "v2",
"created": "Wed, 11 Jun 2025 10:53:13 GMT"
}
] | 2025-06-12T00:00:00 | [
[
"Stan",
"Adriana",
""
],
[
"Combei",
"David",
""
],
[
"Oneata",
"Dan",
""
],
[
"Cucu",
"Horia",
""
]
] |
2506.05816 | Dongzhe Zheng | Dongzhe Zheng | Mirror Symmetry of Spencer-Hodge Decompositions in Constrained Geometric Systems | null | null | null | null | math-ph math.DG math.DS math.MP | http://creativecommons.org/licenses/by/4.0/ | This paper systematically investigates the interaction mechanism between metric structures and mirror transformations in Spencer complexes of compatible pairs. Our core contribution is the establishment of mirror symmetry for Spencer-Hodge decomposition theory, solving the key technical problem of analyzing the behavior of metric geometry under sign transformations. Through precise operator difference analysis, we prove that the perturbation $\mathcal{R}^k = -2(-1)^k \omega \otimes \delta^{\lambda}_{\mathfrak{g}}(s)$ induced by the mirror transformation $(D,\lambda) \mapsto (D,-\lambda)$ is a bounded compact operator, and apply Fredholm theory to establish the mirror invariance of harmonic space dimensions $\dim \mathcal{H}^k_{D,\lambda} = \dim \mathcal{H}^k_{D,-\lambda}$. We further prove the complete invariance of constraint strength metrics and curvature geometric metrics under mirror transformations, thus ensuring the spectral structure stability of Spencer-Hodge Laplacians. From a physical geometric perspective, our results reveal that sign transformations of constraint forces do not affect the essential topological structure of constraint systems, embodying deep symmetry principles in constraint geometry. This work connects Spencer metric theory with mirror symmetry theory, laying the foundation for further development of constraint geometric analysis and computational methods. | [
{
"version": "v1",
"created": "Fri, 06 Jun 2025 07:28:03 GMT"
},
{
"version": "v2",
"created": "Wed, 11 Jun 2025 05:57:48 GMT"
}
] | 2025-06-12T00:00:00 | [
[
"Zheng",
"Dongzhe",
""
]
] |
2506.05915 | Dongzhe Zheng | Dongzhe Zheng | Spencer-Riemann-Roch Theory: Mirror Symmetry of Hodge Decompositions and Characteristic Classes in Constrained Geometry | null | null | null | null | math.AG math.AT math.DG math.RT math.SG | http://creativecommons.org/licenses/by/4.0/ | The discovery of mirror symmetry in compatible pair Spencer complex theory brings new theoretical tools to the study of constrained geometry. Inspired by classical Spencer theory and modern Hodge theory, this paper establishes Spencer-Riemann-Roch theory in the context of constrained geometry, systematically studying the mirror symmetry of Spencer-Hodge decompositions and their manifestations in algebraic geometry. We utilize Serre's GAGA principle to algebraic geometrize Spencer complexes, establish coherent sheaf formulations, and reveal the topological essence of mirror symmetry through characteristic class theory. Main results include: Riemann-Roch type Euler characteristic computation formulas for Spencer complexes, equivalence theorems for mirror symmetry of Hodge decompositions at the characteristic class level, and verification of these theories in concrete geometric constructions. Research shows that algebraic geometric methods can not only reproduce deep results from differential geometry, but also reveal the intrinsic structure of mirror symmetry in constrained geometry through characteristic class analysis, opening new directions for applications of Spencer theory in constrained geometry. | [
{
"version": "v1",
"created": "Fri, 06 Jun 2025 09:36:54 GMT"
},
{
"version": "v2",
"created": "Wed, 11 Jun 2025 06:16:15 GMT"
}
] | 2025-06-12T00:00:00 | [
[
"Zheng",
"Dongzhe",
""
]
] |
2506.05957 | Tianjun Yao | Tianjun Yao, Haoxuan Li, Yongqiang Chen, Tongliang Liu, Le Song, Eric Xing, Zhiqiang Shen | Pruning Spurious Subgraphs for Graph Out-of-Distribtuion Generalization | Submission of ICML2025, with score 4/4/3/3 | null | null | null | cs.LG | http://creativecommons.org/licenses/by/4.0/ | Graph Neural Networks (GNNs) often encounter significant performance degradation under distribution shifts between training and test data, hindering their applicability in real-world scenarios. Recent studies have proposed various methods to address the out-of-distribution generalization challenge, with many methods in the graph domain focusing on directly identifying an invariant subgraph that is predictive of the target label. However, we argue that identifying the edges from the invariant subgraph directly is challenging and error-prone, especially when some spurious edges exhibit strong correlations with the targets. In this paper, we propose PrunE, the first pruning-based graph OOD method that eliminates spurious edges to improve OOD generalizability. By pruning spurious edges, PrunE retains the invariant subgraph more comprehensively, which is critical for OOD generalization. Specifically, PrunE employs two regularization terms to prune spurious edges: 1) graph size constraint to exclude uninformative spurious edges, and 2) $\epsilon$-probability alignment to further suppress the occurrence of spurious edges. Through theoretical analysis and extensive experiments, we show that PrunE achieves superior OOD performance and outperforms previous state-of-the-art methods significantly. Codes are available at: \href{https://github.com/tianyao-aka/PrunE-GraphOOD}{https://github.com/tianyao-aka/PrunE-GraphOOD}. | [
{
"version": "v1",
"created": "Fri, 06 Jun 2025 10:34:48 GMT"
},
{
"version": "v2",
"created": "Tue, 10 Jun 2025 16:58:12 GMT"
},
{
"version": "v3",
"created": "Wed, 11 Jun 2025 12:14:41 GMT"
}
] | 2025-06-12T00:00:00 | [
[
"Yao",
"Tianjun",
""
],
[
"Li",
"Haoxuan",
""
],
[
"Chen",
"Yongqiang",
""
],
[
"Liu",
"Tongliang",
""
],
[
"Song",
"Le",
""
],
[
"Xing",
"Eric",
""
],
[
"Shen",
"Zhiqiang",
""
]
] |
2506.05982 | Zonglin Wu | Zonglin Wu, Yule Xue, Xin Wei, Yiren Song | MCA-Bench: A Multimodal Benchmark for Evaluating CAPTCHA Robustness Against VLM-based Attacks | 31 pages, 8 figures | null | null | null | cs.CV | http://creativecommons.org/licenses/by/4.0/ | As automated attack techniques rapidly advance, CAPTCHAs remain a critical defense mechanism against malicious bots. However, existing CAPTCHA schemes encompass a diverse range of modalities -- from static distorted text and obfuscated images to interactive clicks, sliding puzzles, and logic-based questions -- yet the community still lacks a unified, large-scale, multimodal benchmark to rigorously evaluate their security robustness. To address this gap, we introduce MCA-Bench, a comprehensive and reproducible benchmarking suite that integrates heterogeneous CAPTCHA types into a single evaluation protocol. Leveraging a shared vision-language model backbone, we fine-tune specialized cracking agents for each CAPTCHA category, enabling consistent, cross-modal assessments. Extensive experiments reveal that MCA-Bench effectively maps the vulnerability spectrum of modern CAPTCHA designs under varied attack settings, and crucially offers the first quantitative analysis of how challenge complexity, interaction depth, and model solvability interrelate. Based on these findings, we propose three actionable design principles and identify key open challenges, laying the groundwork for systematic CAPTCHA hardening, fair benchmarking, and broader community collaboration. Datasets and code are available online. | [
{
"version": "v1",
"created": "Fri, 06 Jun 2025 11:02:01 GMT"
},
{
"version": "v2",
"created": "Wed, 11 Jun 2025 07:56:19 GMT"
}
] | 2025-06-12T00:00:00 | [
[
"Wu",
"Zonglin",
""
],
[
"Xue",
"Yule",
""
],
[
"Wei",
"Xin",
""
],
[
"Song",
"Yiren",
""
]
] |
2506.06228 | Ting-Wei Hsu | Ting-Wei Hsu and Hiroyasu Tsukamoto | Statistical Guarantees in Data-Driven Nonlinear Control: Conformal Robustness for Stability and Safety | null | null | 10.1109/LCSYS.2025.3578062 | null | eess.SY cs.SY | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | We present a true-dynamics-agnostic, statistically rigorous framework for establishing exponential stability and safety guarantees of closed-loop, data-driven nonlinear control. Central to our approach is the novel concept of conformal robustness, which robustifies the Lyapunov and zeroing barrier certificates of data-driven dynamical systems against model prediction uncertainties using conformal prediction. It quantifies these uncertainties by leveraging rank statistics of prediction scores over system trajectories, without assuming any specific underlying structure of the prediction model or distribution of the uncertainties. With the quantified uncertainty information, we further construct the conformally robust control Lyapunov function (CR-CLF) and control barrier function (CR-CBF), data-driven counterparts of the CLF and CBF, for fully data-driven control with statistical guarantees of finite-horizon exponential stability and safety. The performance of the proposed concept is validated in numerical simulations with four benchmark nonlinear control problems. | [
{
"version": "v1",
"created": "Fri, 06 Jun 2025 16:42:16 GMT"
}
] | 2025-06-12T00:00:00 | [
[
"Hsu",
"Ting-Wei",
""
],
[
"Tsukamoto",
"Hiroyasu",
""
]
] |
2506.06344 | Joaquin Garcia-Alfaro | Alex Pierron, Michel Barbeau, Luca De Cicco, Jose Rubio-Hernan, Joaquin Garcia-Alfaro | A Reinforcement Learning Approach for RIS-aided Fair Communications | 8 pages, 7 figures, 1 table, 16 references | null | null | null | eess.SP cs.AI cs.LG | http://creativecommons.org/licenses/by/4.0/ | Reconfigurable Intelligent Surfaces (RISs) are composed of physical elements that can dynamically alter electromagnetic wave properties to enhance beamforming and leading to improvements in areas with low coverage properties. They have the potential to be combined with Reinforcement Learning (RL) techniques to achieve network performance and energy efficiency via optimization techniques. In addition to performance and energy improvements, it is also crucial to consider the concept of fair communications. RISs must ensure that User Equipment (UE) units receive their signals with adequate strength, without other UE being deprived of service due to insufficient power. In this paper, we address such a problem. We explore the fairness properties of previous work and propose a novel method that aims at obtaining an efficient and fair duplex RIS-RL system for multiple legitimate UE units. We report and discuss our experimental work and simulation results. We also release our code and datasets to foster further research in the topic. | [
{
"version": "v1",
"created": "Sun, 01 Jun 2025 13:00:26 GMT"
},
{
"version": "v2",
"created": "Wed, 11 Jun 2025 07:31:47 GMT"
}
] | 2025-06-12T00:00:00 | [
[
"Pierron",
"Alex",
""
],
[
"Barbeau",
"Michel",
""
],
[
"De Cicco",
"Luca",
""
],
[
"Rubio-Hernan",
"Jose",
""
],
[
"Garcia-Alfaro",
"Joaquin",
""
]
] |
2506.06395 | Pengyi Li | Pengyi Li, Matvey Skripkin, Alexander Zubrey, Andrey Kuznetsov and Ivan Oseledets | Confidence Is All You Need: Few-Shot RL Fine-Tuning of Language Models | null | null | null | null | cs.CL cs.LG | http://creativecommons.org/licenses/by/4.0/ | Large language models (LLMs) excel at reasoning, yet post-training remains critical for aligning their behavior with task goals. Existing reinforcement learning (RL) methods often depend on costly human annotations or external reward models. We propose Reinforcement Learning via Self-Confidence (RLSC), which uses the model's own confidence as reward signals-eliminating the need for labels, preference models, or reward engineering. Applied to Qwen2.5-Math-7B with only 16 samples per question and 10 or 20 training steps, RLSC improves accuracy by +13.4% on AIME2024, +21.2% on MATH500, +21.7% on Minerva Math, +20.8% on Olympiadbench, and +9.7% on AMC23. RLSC provides a simple, scalable post-training method for inference models, requiring only a small number of samples and unlabelled supervision. | [
{
"version": "v1",
"created": "Thu, 05 Jun 2025 19:55:15 GMT"
},
{
"version": "v2",
"created": "Tue, 10 Jun 2025 14:10:58 GMT"
},
{
"version": "v3",
"created": "Wed, 11 Jun 2025 06:21:59 GMT"
}
] | 2025-06-12T00:00:00 | [
[
"Li",
"Pengyi",
""
],
[
"Skripkin",
"Matvey",
""
],
[
"Zubrey",
"Alexander",
""
],
[
"Kuznetsov",
"Andrey",
""
],
[
"Oseledets",
"Ivan",
""
]
] |
2506.06534 | Leonardo Ermann | Leonardo Ermann, Alexei D. Chepelianskii, Dima L. Shepelyansky | Dynamical thermalization, Rayleigh-Jeans condensate, vortexes and wave collapse in quantum chaos fibers and fluid of light | 35 pages, 33 figures | null | null | null | cond-mat.stat-mech math.DS nlin.CD physics.optics | http://creativecommons.org/licenses/by/4.0/ | We study analytically and numerically the time evolution of a nonlinear field described by the nonlinear Schr\"odinger equation in a chaotic $D$-shape billiard. In absence of nonlinearity the system has standard properties of quantum chaos. This model describes a longitudinal light propagation in a multimode D-shape optical fiber and also those in a Kerr nonlinear medium of atomic vapor. We show that, above a certain chaos border of nonlinearity, chaos leads to dynamical thermalization with the Rayleigh-Jeans thermal distribution and the formation of the Rayleigh-Jeans condensate in a vicinity of the ground state accumulating in it about 80-90\% of total probability. Certain similarities of this phenomenon with the Fr\"ohlich condensate are discussed. Below the chaos border the dynamics is quasi-integrable corresponding to the Kolmogorov-Arnold-Moser integrability. The evolution to the thermal state is characterized by an unusual entropy time dependence with an increase on short times and later significant decrease when approaching to the steady-state. This behavior is opposite to the Boltzmann H-theorem and is attributed to the formation of Rayleigh-Jeans condensate and presence of two integrals of motion, energy and norm. At a strong focusing nonlinearity we show that the wave collapse can take place even at sufficiently high positive energy being very different from the open space case. Finally for the defocusing case we establish the superfluid regime for vortex dynamics at strong nonlinearity.
System parameters for optical fiber experimental studies of these effects are also discussed. | [
{
"version": "v1",
"created": "Fri, 06 Jun 2025 21:05:23 GMT"
},
{
"version": "v2",
"created": "Tue, 10 Jun 2025 20:28:36 GMT"
}
] | 2025-06-12T00:00:00 | [
[
"Ermann",
"Leonardo",
""
],
[
"Chepelianskii",
"Alexei D.",
""
],
[
"Shepelyansky",
"Dima L.",
""
]
] |
2506.06545 | Dominik Wied | Holger Dette, Kathrin M\"ollenhoff, Dominik Wied | Practically significant differences between conditional distribution functions | null | null | null | null | econ.EM stat.ME | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | In the framework of semiparametric distribution regression, we consider the problem of comparing the conditional distribution functions corresponding to two samples. In contrast to testing for exact equality, we are interested in the (null) hypothesis that the $L^2$ distance between the conditional distribution functions does not exceed a certain threshold in absolute value. The consideration of these hypotheses is motivated by the observation that in applications, it is rare, and perhaps impossible, that a null hypothesis of exact equality is satisfied and that the real question of interest is to detect a practically significant deviation between the two conditional distribution functions.
The consideration of a composite null hypothesis makes the testing problem challenging, and in this paper we develop a pivotal test for such hypotheses. Our approach is based on self-normalization and therefore requires neither the estimation of (complicated) variances nor bootstrap approximations. We derive the asymptotic limit distribution of the (appropriately normalized) test statistic and show consistency under local alternatives. A simulation study and an application to German SOEP data reveal the usefulness of the method. | [
{
"version": "v1",
"created": "Fri, 06 Jun 2025 21:25:34 GMT"
}
] | 2025-06-12T00:00:00 | [
[
"Dette",
"Holger",
""
],
[
"Möllenhoff",
"Kathrin",
""
],
[
"Wied",
"Dominik",
""
]
] |
2506.06610 | Dongzhe Zheng | Dongzhe Zheng | Mirror Symmetry in Geometric Constraints: Analytic and Riemann-Roch Perspectives | null | null | null | null | math.AG math-ph math.DG math.MP | http://creativecommons.org/licenses/by/4.0/ | Building upon the discovery of mirror symmetry phenomena in compatible pair Spencer complexes, this paper develops rigorous analytical foundations through systematic application of elliptic perturbation theory and characteristic class analysis. We establish a comprehensive multi-level framework revealing the mathematical mechanisms underlying the symbolic transformation $(D,\lambda) \mapsto (D,-\lambda)$: metric invariance analysis via Spencer metric theory, topological isomorphisms through elliptic operator perturbation methods, and algebraic geometric equivalences via GAGA principle and characteristic class machinery.
Main results include: (1) Rigorous proof of Spencer-Hodge Laplacian perturbation relations and cohomological mirror isomorphisms using Fredholm theory; (2) Complete characteristic class equivalence theorems establishing algebraic geometric foundations for mirror symmetry; (3) Systematic Spencer-Riemann-Roch decomposition formulas with explicit error estimates on Calabi-Yau manifolds; (4) Computational verification frameworks demonstrated on elliptic curves.
This research demonstrates that mirror symmetry in constraint geometry possesses deep analytical structure accessible through modern elliptic theory and algebraic geometry, establishing Spencer complexes as a bridge between constraint analysis and contemporary geometric theory. | [
{
"version": "v1",
"created": "Sat, 07 Jun 2025 01:03:30 GMT"
},
{
"version": "v2",
"created": "Wed, 11 Jun 2025 06:30:59 GMT"
}
] | 2025-06-12T00:00:00 | [
[
"Zheng",
"Dongzhe",
""
]
] |
2506.06671 | Pedro Teles Dr | Pedro Teles | An analytical model for gold nanoparticle radiosensitisation | null | null | null | null | physics.med-ph | http://creativecommons.org/licenses/by-sa/4.0/ | In this paper, we derive a variance-driven Local-Effect-Model ($\sigma$-LEM) to predict radiosensitization due to gold nanoparticles (AuNP). Assuming that the number of Au photo-ionisations scales strictly with particle volume $V_{\mathrm{NP}} \propto R^{3}$ and that the cascade energy deposition is log-normally distributed, the enhanced dose in each target voxel can be written as $D_{\text{enh}} = D_{0} \exp(\sigma Z)$ with $Z \sim \mathcal{N}(0,1)$ and width $\sigma = \sqrt{2 \ln(1 + Kc)}$.
By assuming a linear-quadratic (LQ) dose response, a relation between survival and dose can be derived for each voxel. Although there is no closed form for the log-normal distribution, averaging over the entire domain using first and second order moments leads to three possible closed forms: a variance-only one, a mixed-term one, where simultaneous hits from enhancement and baseline can occur, and a second-order model, accurate to $\mathcal{O}(\sigma^{3})$. These three variants adapt well to low-concentration, mid-concentration, and high-concentration regimes.
The model was tested for Bovine aortic endothelial cells (BAEC) results, irradiated with a 100 kVp x-ray beam, taken from a Local Effect Model (LEM) and experimental values. The model shows good agreement with the experimental data, with deviations within acceptable limits, but presents significant changes to the results obtained with the LEM. In particular, results seem to indicate that dose enhancement is mostly $\alpha$-driven. These findings are further developed in the manuscript.
The theoretical framework presented here collapses radiobiological outcomes to two experimentally controllable variables, beam quality, in the term $K_c$, and intracellular concentration $c$, while retaining mechanistic fidelity. Additional tests should be made to further confirm the validity of the model. | [
{
"version": "v1",
"created": "Sat, 07 Jun 2025 05:28:46 GMT"
},
{
"version": "v2",
"created": "Wed, 11 Jun 2025 02:12:03 GMT"
}
] | 2025-06-12T00:00:00 | [
[
"Teles",
"Pedro",
""
]
] |
2506.06733 | Ruoxuan Zhang | Ruoxuan Zhang, Jidong Gao, Bin Wen, Hongxia Xie, Chenming Zhang, Hong-Han Shuai, Wen-Huang Cheng | RecipeGen: A Step-Aligned Multimodal Benchmark for Real-World Recipe Generation | This is an extended version of arXiv:2503.05228 | null | null | null | cs.CV | http://creativecommons.org/licenses/by-nc-nd/4.0/ | Creating recipe images is a key challenge in food computing, with applications in culinary education and multimodal recipe assistants. However, existing datasets lack fine-grained alignment between recipe goals, step-wise instructions, and visual content. We present RecipeGen, the first large-scale, real-world benchmark for recipe-based Text-to-Image (T2I), Image-to-Video (I2V), and Text-to-Video (T2V) generation. RecipeGen contains 26,453 recipes, 196,724 images, and 4,491 videos, covering diverse ingredients, cooking procedures, styles, and dish types. We further propose domain-specific evaluation metrics to assess ingredient fidelity and interaction modeling, benchmark representative T2I, I2V, and T2V models, and provide insights for future recipe generation models. Project page is available now. | [
{
"version": "v1",
"created": "Sat, 07 Jun 2025 09:35:35 GMT"
},
{
"version": "v2",
"created": "Tue, 10 Jun 2025 09:24:29 GMT"
},
{
"version": "v3",
"created": "Wed, 11 Jun 2025 02:08:22 GMT"
}
] | 2025-06-12T00:00:00 | [
[
"Zhang",
"Ruoxuan",
""
],
[
"Gao",
"Jidong",
""
],
[
"Wen",
"Bin",
""
],
[
"Xie",
"Hongxia",
""
],
[
"Zhang",
"Chenming",
""
],
[
"Shuai",
"Hong-Han",
""
],
[
"Cheng",
"Wen-Huang",
""
]
] |
2506.06745 | Morteza Jazandari | Morteza Jazandari, Jahanfar Abouie, and Daryoosh Vashaee | What Really Drives Thermopower: Specific Heat or Entropy as the Unifying Principle Across Magnetic, Superconducting, and Nanoscale Systems | 23 pages, 11 figures | null | null | null | cond-mat.mtrl-sci | http://creativecommons.org/licenses/by/4.0/ | Thermopower, a key parameter in thermoelectric performance, is often linked to either specific heat or entropy, yet the fundamental quantity that governs it has remained elusive. In this work, we present a unified theoretical framework that identifies entropy per carrier, not specific heat, as the universal driver of thermopower across both closed and open systems. Using thermodynamic identities and the Onsager-Kelvin relation, we show that thermopower is universally proportional to entropy per carrier, while its apparent proportionality to specific heat arises only in systems where the specific heat follows a continuous power-law temperature dependence. To extend this framework to magnetic systems, we derive a general expression for magnon-drag thermopower that holds in both Newtonian (massive, parabolic) and relativistic (massless, linear) magnon regimes. In particular, we reformulate the momentum balance using a relativistic energy-momentum tensor, resolving conceptual inconsistencies in prior models that relied on ill-defined magnon masses in antiferromagnets. Our framework is further illustrated through three representative systems: (i) magnetic materials, where magnon and paramagnon entropy sustain thermopower across TC and TN; (ii) superconducting Nb, where anomalous thermopower emerges from entropy carried by Bogoliubov quasiparticles near TC; and (iii) a single-molecule junction, where entropy from occupation-number fluctuations governs thermopower in an open quantum system. We validate our unifying principle by comparing it with experimental data: thermopower measurements of superconducting niobium reveal the role of quasiparticle entropy near the critical temperature, and literature-reported specific heat data from a wide range of ferromagnetic and antiferromagnetic materials demonstrate consistent entropy-based scaling across magnetic transitions. | [
{
"version": "v1",
"created": "Sat, 07 Jun 2025 10:25:33 GMT"
},
{
"version": "v2",
"created": "Tue, 10 Jun 2025 19:13:36 GMT"
}
] | 2025-06-12T00:00:00 | [
[
"Jazandari",
"Morteza",
""
],
[
"Abouie",
"Jahanfar",
""
],
[
"Vashaee",
"Daryoosh",
""
]
] |
2506.06783 | David Maibach | David Maibach | Across the Horizon: On Gravitational Wave Flux Laws and Tests of Gravity | 374 pages, Ph.D. Thesis, Heidelberg, University of Heidelberg (2025) | null | null | null | gr-qc | http://creativecommons.org/licenses/by/4.0/ | Motivated by the first detection of gravitational waves, this dissertation develops analytical, numerical, and data analysis techniques to address persistent blind spots in our understanding of gravity. Beginning with asymptotically flat spacetimes and the geometry of null geodesic congruences, the derivation of the shear tensor--encoding gravitational radiation--is revisited. The covariant phase space formulation of General Relativity is then employed to derive a non-conservation law associated with the symmetry group at null infinity, generalizing prior constructions of radiative phase space and yielding consistent results. These flux laws are used to derive constraint equations that enable the evaluation and comparison of state-of-the-art numerical waveform models, leading to new insights and a robust algorithm for benchmarking future improvements. These flux laws are further applied to compute quantum corrections to gravitational waveforms arising from the gravitational wave echo effect. Two leading phenomenological scenarios involving echoes from binary black hole mergers are analyzed. The structure of both the primary echo signal and its induced corrections to nonlinear features of the waveform are studied for observability by the upcoming LISA mission. The results indicate that LISA could detect such signals, potentially probing black hole area quantization. Finally, motivated by recent hints of a stochastic gravitational wave background from Pulsar Timing Array data, this work reviews its theoretical basis and studies several astrophysical and cosmological source scenarios. A forecast for detection prospects with LISA is presented using a modern data analysis pipeline. The results suggest LISA will constrain the extra-galactic stochastic background to a spectral energy density below $ \Omega_\text{GW} \lesssim 10^{-8}$. | [
{
"version": "v1",
"created": "Sat, 07 Jun 2025 12:51:38 GMT"
},
{
"version": "v2",
"created": "Wed, 11 Jun 2025 12:33:03 GMT"
}
] | 2025-06-12T00:00:00 | [
[
"Maibach",
"David",
""
]
] |
2506.06821 | Yuhan Cao | Yuhan Cao, Zian Chen, Kun Quan, Ziliang Zhang, Yu Wang, Xiaoning Dong, Yeqi Feng, Guanzhong He, Jingcheng Huang, Jianhao Li, Yixuan Tan, Jiafu Tang, Yilin Tang, Junlei Wu, Qianyu Xiao, Can Zheng, Shouchen Zhou, Yuxiang Zhu, Yiming Huang, Tian Xie, Tianxing He | Can LLMs Generate Reliable Test Case Generators? A Study on Competition-Level Programming Problems | 37 pages, 22 figures | null | null | null | cs.CL cs.AI cs.SE | http://creativecommons.org/licenses/by-nc-nd/4.0/ | Large Language Models (LLMs) have demonstrated remarkable capabilities in code generation, capable of tackling complex tasks during inference. However, the extent to which LLMs can be utilized for code checking or debugging through test case generation remains largely unexplored. We investigate this problem from the perspective of competition-level programming (CP) programs and propose TCGBench, a Benchmark for (LLM generation of) Test Case Generators. This benchmark comprises two tasks, aimed at studying the capabilities of LLMs in (1) generating valid test case generators for a given CP problem, and further (2) generating targeted test case generators that expose bugs in human-written code. Experimental results indicate that while state-of-the-art LLMs can generate valid test case generators in most cases, most LLMs struggle to generate targeted test cases that reveal flaws in human code effectively. Especially, even advanced reasoning models (e.g., o3-mini) fall significantly short of human performance in the task of generating targeted generators. Furthermore, we construct a high-quality, manually curated dataset of instructions for generating targeted generators. Analysis demonstrates that the performance of LLMs can be enhanced with the aid of this dataset, by both prompting and fine-tuning. | [
{
"version": "v1",
"created": "Sat, 07 Jun 2025 14:53:03 GMT"
},
{
"version": "v2",
"created": "Tue, 10 Jun 2025 19:38:57 GMT"
}
] | 2025-06-12T00:00:00 | [
[
"Cao",
"Yuhan",
""
],
[
"Chen",
"Zian",
""
],
[
"Quan",
"Kun",
""
],
[
"Zhang",
"Ziliang",
""
],
[
"Wang",
"Yu",
""
],
[
"Dong",
"Xiaoning",
""
],
[
"Feng",
"Yeqi",
""
],
[
"He",
"Guanzhong",
""
],
[
"Huang",
"Jingcheng",
""
],
[
"Li",
"Jianhao",
""
],
[
"Tan",
"Yixuan",
""
],
[
"Tang",
"Jiafu",
""
],
[
"Tang",
"Yilin",
""
],
[
"Wu",
"Junlei",
""
],
[
"Xiao",
"Qianyu",
""
],
[
"Zheng",
"Can",
""
],
[
"Zhou",
"Shouchen",
""
],
[
"Zhu",
"Yuxiang",
""
],
[
"Huang",
"Yiming",
""
],
[
"Xie",
"Tian",
""
],
[
"He",
"Tianxing",
""
]
] |
2506.06848 | D. Carlos Akkar | D. Carlos Akkar | The (Mis)use of Information in Decentralised Markets | 61 pages, 9 figures | null | null | null | econ.TH | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | A seller offers an asset in a decentralised market. Buyers have private signals about their common value. I study whether the market becomes allocatively more efficient with (i) more buyers, (ii) better-informed buyers. Both increase the information available about buyers' common value, but also the adverse selection each buyer faces. With more buyers, trade surplus eventually increases and converges to the full-information upper bound if and only if the likelihood ratios of buyers' signals are unbounded from above. Otherwise, it eventually decreases and converges to the no-information lower bound. With better information about trades buyers would have accepted, trade surplus increases. With better information about trades they would have rejected, trade surplus decreases--unless adverse selection is irrelevant. For binary signals, a sharper characterisation emerges: stronger good news increase total surplus, but stronger bad news eventually decrease it. | [
{
"version": "v1",
"created": "Sat, 07 Jun 2025 16:00:52 GMT"
},
{
"version": "v2",
"created": "Tue, 10 Jun 2025 09:32:32 GMT"
},
{
"version": "v3",
"created": "Wed, 11 Jun 2025 09:33:18 GMT"
}
] | 2025-06-12T00:00:00 | [
[
"Akkar",
"D. Carlos",
""
]
] |
2506.06905 | Akash Gupta | Akash Gupta, Amos Storkey, Mirella Lapata | Meta-Adaptive Prompt Distillation for Few-Shot Visual Question Answering | null | null | null | null | cs.AI cs.CL cs.CV cs.LG | http://creativecommons.org/licenses/by/4.0/ | Large Multimodal Models (LMMs) often rely on in-context learning (ICL) to perform new tasks with minimal supervision. However, ICL performance, especially in smaller LMMs, is inconsistent and does not always improve monotonically with increasing examples. We hypothesize that this occurs due to the LMM being overwhelmed by additional information present in the image embeddings, which is not required for the downstream task. To address this, we propose a meta-learning approach that provides an alternative for inducing few-shot capabilities in LMMs, using a fixed set of soft prompts that are distilled from task-relevant image features and can be adapted at test time using a few examples. To facilitate this distillation, we introduce an attention-mapper module that can be easily integrated with the popular LLaVA v1.5 architecture and is jointly learned with soft prompts, enabling task adaptation in LMMs under low-data regimes with just a few gradient steps. Evaluation on the VL-ICL Bench shows that our method consistently outperforms ICL and related prompt-tuning approaches, even under image perturbations, improving task induction and reasoning across visual question answering tasks. | [
{
"version": "v1",
"created": "Sat, 07 Jun 2025 19:37:22 GMT"
},
{
"version": "v2",
"created": "Tue, 10 Jun 2025 07:34:44 GMT"
}
] | 2025-06-12T00:00:00 | [
[
"Gupta",
"Akash",
""
],
[
"Storkey",
"Amos",
""
],
[
"Lapata",
"Mirella",
""
]
] |
2506.06975 | Xiaoyuan Zhu | Xiaoyuan Zhu, Yaowen Ye, Tianyi Qiu, Hanlin Zhu, Sijun Tan, Ajraf Mannan, Jonathan Michala, Raluca Ada Popa, Willie Neiswanger | Auditing Black-Box LLM APIs with a Rank-Based Uniformity Test | null | null | null | null | cs.CR cs.AI cs.CL | http://creativecommons.org/licenses/by/4.0/ | As API access becomes a primary interface to large language models (LLMs), users often interact with black-box systems that offer little transparency into the deployed model. To reduce costs or maliciously alter model behaviors, API providers may discreetly serve quantized or fine-tuned variants, which can degrade performance and compromise safety. Detecting such substitutions is difficult, as users lack access to model weights and, in most cases, even output logits. To tackle this problem, we propose a rank-based uniformity test that can verify the behavioral equality of a black-box LLM to a locally deployed authentic model. Our method is accurate, query-efficient, and avoids detectable query patterns, making it robust to adversarial providers that reroute or mix responses upon the detection of testing attempts. We evaluate the approach across diverse threat scenarios, including quantization, harmful fine-tuning, jailbreak prompts, and full model substitution, showing that it consistently achieves superior statistical power over prior methods under constrained query budgets. | [
{
"version": "v1",
"created": "Sun, 08 Jun 2025 03:00:31 GMT"
},
{
"version": "v2",
"created": "Tue, 10 Jun 2025 02:21:06 GMT"
},
{
"version": "v3",
"created": "Wed, 11 Jun 2025 03:04:10 GMT"
}
] | 2025-06-12T00:00:00 | [
[
"Zhu",
"Xiaoyuan",
""
],
[
"Ye",
"Yaowen",
""
],
[
"Qiu",
"Tianyi",
""
],
[
"Zhu",
"Hanlin",
""
],
[
"Tan",
"Sijun",
""
],
[
"Mannan",
"Ajraf",
""
],
[
"Michala",
"Jonathan",
""
],
[
"Popa",
"Raluca Ada",
""
],
[
"Neiswanger",
"Willie",
""
]
] |
2506.06985 | Anastasiia Koloskova | Anastasia Koloskova, Youssef Allouah, Animesh Jha, Rachid Guerraoui, Sanmi Koyejo | Certified Unlearning for Neural Networks | null | null | null | null | cs.LG cs.CR stat.ML | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | We address the problem of machine unlearning, where the goal is to remove the influence of specific training data from a model upon request, motivated by privacy concerns and regulatory requirements such as the "right to be forgotten." Unfortunately, existing methods rely on restrictive assumptions or lack formal guarantees. To this end, we propose a novel method for certified machine unlearning, leveraging the connection between unlearning and privacy amplification by stochastic post-processing. Our method uses noisy fine-tuning on the retain data, i.e., data that does not need to be removed, to ensure provable unlearning guarantees. This approach requires no assumptions about the underlying loss function, making it broadly applicable across diverse settings. We analyze the theoretical trade-offs in efficiency and accuracy and demonstrate empirically that our method not only achieves formal unlearning guarantees but also performs effectively in practice, outperforming existing baselines. Our code is available at https://github.com/stair-lab/certified-unlearning-neural-networks-icml-2025 | [
{
"version": "v1",
"created": "Sun, 08 Jun 2025 03:55:28 GMT"
},
{
"version": "v2",
"created": "Tue, 10 Jun 2025 19:29:50 GMT"
}
] | 2025-06-12T00:00:00 | [
[
"Koloskova",
"Anastasia",
""
],
[
"Allouah",
"Youssef",
""
],
[
"Jha",
"Animesh",
""
],
[
"Guerraoui",
"Rachid",
""
],
[
"Koyejo",
"Sanmi",
""
]
] |
2506.06990 | Mingyi Li | Mingyi Li, Michael R. Metel, Akiko Takeda | Modified K-means Algorithm with Local Optimality Guarantees | ICML 2025 | null | null | null | cs.LG math.OC stat.ML | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | The K-means algorithm is one of the most widely studied clustering algorithms in machine learning. While extensive research has focused on its ability to achieve a globally optimal solution, there still lacks a rigorous analysis of its local optimality guarantees. In this paper, we first present conditions under which the K-means algorithm converges to a locally optimal solution. Based on this, we propose simple modifications to the K-means algorithm which ensure local optimality in both the continuous and discrete sense, with the same computational complexity as the original K-means algorithm. As the dissimilarity measure, we consider a general Bregman divergence, which is an extension of the squared Euclidean distance often used in the K-means algorithm. Numerical experiments confirm that the K-means algorithm does not always find a locally optimal solution in practice, while our proposed methods provide improved locally optimal solutions with reduced clustering loss. Our code is available at https://github.com/lmingyi/LO-K-means. | [
{
"version": "v1",
"created": "Sun, 08 Jun 2025 04:37:28 GMT"
},
{
"version": "v2",
"created": "Wed, 11 Jun 2025 06:52:53 GMT"
}
] | 2025-06-12T00:00:00 | [
[
"Li",
"Mingyi",
""
],
[
"Metel",
"Michael R.",
""
],
[
"Takeda",
"Akiko",
""
]
] |
2506.07044 | Weiwen Xu | LASA Team, Weiwen Xu, Hou Pong Chan, Long Li, Mahani Aljunied, Ruifeng Yuan, Jianyu Wang, Chenghao Xiao, Guizhen Chen, Chaoqun Liu, Zhaodonghui Li, Yu Sun, Junao Shen, Chaojun Wang, Jie Tan, Deli Zhao, Tingyang Xu, Hao Zhang, Yu Rong | Lingshu: A Generalist Foundation Model for Unified Multimodal Medical Understanding and Reasoning | Technical Report, 53 pages, 25 tables, and 16 figures | null | null | null | cs.CL cs.AI cs.CV | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Multimodal Large Language Models (MLLMs) have demonstrated impressive capabilities in understanding common visual elements, largely due to their large-scale datasets and advanced training strategies. However, their effectiveness in medical applications remains limited due to the inherent discrepancies between data and tasks in medical scenarios and those in the general domain. Concretely, existing medical MLLMs face the following critical limitations: (1) limited coverage of medical knowledge beyond imaging, (2) heightened susceptibility to hallucinations due to suboptimal data curation processes, (3) lack of reasoning capabilities tailored for complex medical scenarios. To address these challenges, we first propose a comprehensive data curation procedure that (1) efficiently acquires rich medical knowledge data not only from medical imaging but also from extensive medical texts and general-domain data; and (2) synthesizes accurate medical captions, visual question answering (VQA), and reasoning samples. As a result, we build a multimodal dataset enriched with extensive medical knowledge. Building on the curated data, we introduce our medical-specialized MLLM: Lingshu. Lingshu undergoes multi-stage training to embed medical expertise and enhance its task-solving capabilities progressively. Besides, we preliminarily explore the potential of applying reinforcement learning with verifiable rewards paradigm to enhance Lingshu's medical reasoning ability. Additionally, we develop MedEvalKit, a unified evaluation framework that consolidates leading multimodal and textual medical benchmarks for standardized, fair, and efficient model assessment. We evaluate the performance of Lingshu on three fundamental medical tasks, multimodal QA, text-based QA, and medical report generation. The results show that Lingshu consistently outperforms the existing open-source multimodal models on most tasks ... | [
{
"version": "v1",
"created": "Sun, 08 Jun 2025 08:47:30 GMT"
},
{
"version": "v2",
"created": "Tue, 10 Jun 2025 07:38:24 GMT"
},
{
"version": "v3",
"created": "Wed, 11 Jun 2025 09:12:18 GMT"
}
] | 2025-06-12T00:00:00 | [
[
"LASA Team",
"",
""
],
[
"Xu",
"Weiwen",
""
],
[
"Chan",
"Hou Pong",
""
],
[
"Li",
"Long",
""
],
[
"Aljunied",
"Mahani",
""
],
[
"Yuan",
"Ruifeng",
""
],
[
"Wang",
"Jianyu",
""
],
[
"Xiao",
"Chenghao",
""
],
[
"Chen",
"Guizhen",
""
],
[
"Liu",
"Chaoqun",
""
],
[
"Li",
"Zhaodonghui",
""
],
[
"Sun",
"Yu",
""
],
[
"Shen",
"Junao",
""
],
[
"Wang",
"Chaojun",
""
],
[
"Tan",
"Jie",
""
],
[
"Zhao",
"Deli",
""
],
[
"Xu",
"Tingyang",
""
],
[
"Zhang",
"Hao",
""
],
[
"Rong",
"Yu",
""
]
] |
2506.07088 | Ilja Kuzborskij | Ilja Kuzborskij, Yasin Abbasi Yadkori | Pointwise confidence estimation in the non-linear $\ell^2$-regularized least squares | null | null | null | null | cs.LG stat.ML | http://creativecommons.org/licenses/by/4.0/ | We consider a high-probability non-asymptotic confidence estimation in the $\ell^2$-regularized non-linear least-squares setting with fixed design. In particular, we study confidence estimation for local minimizers of the regularized training loss. We show a pointwise confidence bound, meaning that it holds for the prediction on any given fixed test input $x$. Importantly, the proposed confidence bound scales with similarity of the test input to the training data in the implicit feature space of the predictor (for instance, becoming very large when the test input lies far outside of the training data). This desirable last feature is captured by the weighted norm involving the inverse-Hessian matrix of the objective function, which is a generalized version of its counterpart in the linear setting, $x^{\top} \text{Cov}^{-1} x$. Our generalized result can be regarded as a non-asymptotic counterpart of the classical confidence interval based on asymptotic normality of the MLE estimator. We propose an efficient method for computing the weighted norm, which only mildly exceeds the cost of a gradient computation of the loss function. Finally, we complement our analysis with empirical evidence showing that the proposed confidence bound provides better coverage/width trade-off compared to a confidence estimation by bootstrapping, which is a gold-standard method in many applications involving non-linear predictors such as neural networks. | [
{
"version": "v1",
"created": "Sun, 08 Jun 2025 11:23:49 GMT"
},
{
"version": "v2",
"created": "Tue, 10 Jun 2025 18:59:00 GMT"
}
] | 2025-06-12T00:00:00 | [
[
"Kuzborskij",
"Ilja",
""
],
[
"Yadkori",
"Yasin Abbasi",
""
]
] |
2506.07206 | Fengyi Song | Fengyi Song, Decai Liang, Changliang Zou | Change-Points Detection and Support Recovery for Spatially Indexed Functional Data | null | null | null | null | stat.ME | http://creativecommons.org/licenses/by-nc-sa/4.0/ | Large volumes of spatiotemporal data, characterized by high spatial and temporal variability, may experience structural changes over time. Unlike traditional change-point problems, each sequence in this context consists of function-valued curves observed at multiple spatial locations, with typically only a small subset of locations affected. This paper addresses two key issues: detecting the global change-point and identifying the spatial support set, within a unified framework tailored to spatially indexed functional data. By leveraging a weakly separable cross-covariance structure -- an extension beyond the restrictive assumption of space-time separability -- we incorporate functional principal component analysis into the change-detection methodology, while preserving common temporal features across locations. A kernel-based test statistic is further developed to integrate spatial clustering pattern into the detection process, and its local variant, combined with the estimated change-point, is employed to identify the subset of locations contributing to the mean shifts. To control the false discovery rate in multiple testing, we introduce a functional symmetrized data aggregation approach that does not rely on pointwise p-values and effectively pools spatial information. We establish the asymptotic validity of the proposed change detection and support recovery method under mild regularity conditions. The efficacy of our approach is demonstrated through simulations, with its practical usefulness illustrated in an application to China's precipitation data. | [
{
"version": "v1",
"created": "Sun, 08 Jun 2025 16:12:37 GMT"
},
{
"version": "v2",
"created": "Wed, 11 Jun 2025 01:59:33 GMT"
}
] | 2025-06-12T00:00:00 | [
[
"Song",
"Fengyi",
""
],
[
"Liang",
"Decai",
""
],
[
"Zou",
"Changliang",
""
]
] |
2506.07237 | Jui-Chiang Wei | Jui-Chiang Wei, Yi-Cheng Lin, Fabian Ritter-Gutierrez, Hung-yi Lee | Multi-Distillation from Speech and Music Representation Models | 8 pages, 1 figures | null | null | null | eess.AS | http://creativecommons.org/licenses/by/4.0/ | Real-world audio often mixes speech and music, yet models typically handle only one domain. This paper introduces a multi-teacher distillation framework that unifies speech and music models into a single one while significantly reducing model size. Our approach leverages the strengths of domain-specific teacher models, such as HuBERT for speech and MERT for music, and explores various strategies to balance both domains. Experiments across diverse tasks demonstrate that our model matches the performance of domain-specific models, showing the effectiveness of cross-domain distillation. Additionally, we conduct few-shot learning experiments, highlighting the need for general models in real-world scenarios where labeled data is limited. Our results show that our model not only performs on par with specialized models but also outperforms them in few-shot scenarios, proving that a cross-domain approach is essential and effective for diverse tasks with limited data. | [
{
"version": "v1",
"created": "Sun, 08 Jun 2025 17:45:46 GMT"
},
{
"version": "v2",
"created": "Wed, 11 Jun 2025 17:28:22 GMT"
}
] | 2025-06-12T00:00:00 | [
[
"Wei",
"Jui-Chiang",
""
],
[
"Lin",
"Yi-Cheng",
""
],
[
"Ritter-Gutierrez",
"Fabian",
""
],
[
"Lee",
"Hung-yi",
""
]
] |
2506.07254 | Kevin Frans | Kevin Frans, Sergey Levine, Pieter Abbeel | A Stable Whitening Optimizer for Efficient Neural Network Training | null | null | null | null | cs.LG | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | In this work, we take an experimentally grounded look at neural network optimization. Building on the Shampoo family of algorithms, we identify and alleviate three key issues, resulting in the proposed SPlus method. First, we find that naive Shampoo is prone to divergence when matrix-inverses are cached for long periods. We introduce an alternate bounded update combining a historical eigenbasis with instantaneous normalization, resulting in across-the-board stability and significantly lower computational requirements. Second, we adapt a shape-aware scaling to enable learning rate transfer across network width. Third, we find that high learning rates result in large parameter noise, and propose a simple iterate-averaging scheme which unblocks faster learning. To properly confirm these findings, we introduce a pointed Transformer training benchmark, considering three objectives (language modelling, image classification, and diffusion modelling) across different stages of training. On average, SPlus is able to reach the validation performance of Adam within 44% of the gradient steps and 62% of the wallclock time. | [
{
"version": "v1",
"created": "Sun, 08 Jun 2025 18:43:31 GMT"
},
{
"version": "v2",
"created": "Tue, 10 Jun 2025 22:01:14 GMT"
}
] | 2025-06-12T00:00:00 | [
[
"Frans",
"Kevin",
""
],
[
"Levine",
"Sergey",
""
],
[
"Abbeel",
"Pieter",
""
]
] |
2506.07288 | Weijie Guan | Weijie Guan, Haohui Wang, Jian Kang, Lihui Liu, and Dawei Zhou | EVINET: Towards Open-World Graph Learning via Evidential Reasoning Network | KDD 2025 | null | 10.1145/3711896.3736945 | null | cs.LG | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Graph learning has been crucial to many real-world tasks, but they are often studied with a closed-world assumption, with all possible labels of data known a priori. To enable effective graph learning in an open and noisy environment, it is critical to inform the model users when the model makes a wrong prediction to in-distribution data of a known class, i.e., misclassification detection or when the model encounters out-of-distribution from novel classes, i.e., out-of-distribution detection. This paper introduces Evidential Reasoning Network (EVINET), a framework that addresses these two challenges by integrating Beta embedding within a subjective logic framework. EVINET includes two key modules: Dissonance Reasoning for misclassification detection and Vacuity Reasoning for out-of-distribution detection. Extensive experiments demonstrate that EVINET outperforms state-of-the-art methods across multiple metrics in the tasks of in-distribution classification, misclassification detection, and out-of-distribution detection. EVINET demonstrates the necessity of uncertainty estimation and logical reasoning for misclassification detection and out-of-distribution detection and paves the way for open-world graph learning. Our code and data are available at https://github.com/SSSKJ/EviNET. | [
{
"version": "v1",
"created": "Sun, 08 Jun 2025 21:21:12 GMT"
},
{
"version": "v2",
"created": "Wed, 11 Jun 2025 17:59:50 GMT"
}
] | 2025-06-12T00:00:00 | [
[
"Guan",
"Weijie",
""
],
[
"Wang",
"Haohui",
""
],
[
"Kang",
"Jian",
""
],
[
"Liu",
"Lihui",
""
],
[
"Zhou",
"Dawei",
""
]
] |
2506.07298 | Yijia Dai | Yijia Dai, Zhaolin Gao, Yahya Sattar, Sarah Dean, Jennifer J. Sun | Pre-trained Large Language Models Learn Hidden Markov Models In-context | null | null | null | null | cs.LG cs.AI | http://creativecommons.org/licenses/by/4.0/ | Hidden Markov Models (HMMs) are foundational tools for modeling sequential data with latent Markovian structure, yet fitting them to real-world data remains computationally challenging. In this work, we show that pre-trained large language models (LLMs) can effectively model data generated by HMMs via in-context learning (ICL)$\unicode{x2013}$their ability to infer patterns from examples within a prompt. On a diverse set of synthetic HMMs, LLMs achieve predictive accuracy approaching the theoretical optimum. We uncover novel scaling trends influenced by HMM properties, and offer theoretical conjectures for these empirical observations. We also provide practical guidelines for scientists on using ICL as a diagnostic tool for complex data. On real-world animal decision-making tasks, ICL achieves competitive performance with models designed by human experts. To our knowledge, this is the first demonstration that ICL can learn and predict HMM-generated sequences$\unicode{x2013}$an advance that deepens our understanding of in-context learning in LLMs and establishes its potential as a powerful tool for uncovering hidden structure in complex scientific data. | [
{
"version": "v1",
"created": "Sun, 08 Jun 2025 21:49:38 GMT"
},
{
"version": "v2",
"created": "Wed, 11 Jun 2025 05:17:22 GMT"
}
] | 2025-06-12T00:00:00 | [
[
"Dai",
"Yijia",
""
],
[
"Gao",
"Zhaolin",
""
],
[
"Sattar",
"Yahya",
""
],
[
"Dean",
"Sarah",
""
],
[
"Sun",
"Jennifer J.",
""
]
] |
2506.07300 | Thodoris Spanos | Thodoris Spanos, Fran Fabra, Jos\'e A. L\'opez-Salcedo, Gonzalo Seco-Granados, Nikos Kanistras, Ivan Lapin, Vassilis Paliouras | Theoretical and Experimental Evaluation of AoA Estimation in Single-Anchor 5G Uplink Positioning | 6 pages, 6 figures, 2 tables Accepted for publication in the 2025 IEEE International Symposium on Personal, Indoor and Mobile Radio Communications (PIMRC), Istanbul, Turkey, September 1-4, 2025; to appear in IEEE PIMRC 2025 proceedings. \c{opyright} 2025 IEEE. Personal use of this material is permitted | null | null | null | eess.SP | http://creativecommons.org/licenses/by/4.0/ | As we move towards 6G, the demand for high-precision, cost-effective positioning solutions becomes increasingly critical. Single-anchor positioning offers a promising alternative to traditional multi-anchor approaches, particularly in complex propagation environments where infrastructure costs and deployment constraints present significant challenges. This paper provides a comprehensive evaluation of key algorithmic choices in the development of a single-anchor 5G uplink positioning testbed. Our developed testbed uses angle of arrival (AoA) estimation combined with range measurements from an ultra-wideband pair, to derive the position. The simulations conducted assess the impact of the selected algorithms on channel order and AoA estimation, while the influence of antenna calibration errors on AoA estimation is also examined. Finally, we compare simulations and results obtained from our developed platform. | [
{
"version": "v1",
"created": "Sun, 08 Jun 2025 21:56:26 GMT"
},
{
"version": "v2",
"created": "Tue, 10 Jun 2025 11:44:50 GMT"
},
{
"version": "v3",
"created": "Wed, 11 Jun 2025 08:08:58 GMT"
}
] | 2025-06-12T00:00:00 | [
[
"Spanos",
"Thodoris",
""
],
[
"Fabra",
"Fran",
""
],
[
"López-Salcedo",
"José A.",
""
],
[
"Seco-Granados",
"Gonzalo",
""
],
[
"Kanistras",
"Nikos",
""
],
[
"Lapin",
"Ivan",
""
],
[
"Paliouras",
"Vassilis",
""
]
] |
2506.07400 | Philip Liu | Philip R. Liu, Sparsh Bansal, Jimmy Dinh, Aditya Pawar, Ramani Satishkumar, Shail Desai, Neeraj Gupta, Xin Wang, Shu Hu | MedChat: A Multi-Agent Framework for Multimodal Diagnosis with Large Language Models | 7 pages, 6 figures. Accepted to the 2025 IEEE 8th International Conference on Multimedia Information Processing and Retrieval (MIPR) | null | null | null | cs.MA cs.AI cs.CV cs.LG | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | The integration of deep learning-based glaucoma detection with large language models (LLMs) presents an automated strategy to mitigate ophthalmologist shortages and improve clinical reporting efficiency. However, applying general LLMs to medical imaging remains challenging due to hallucinations, limited interpretability, and insufficient domain-specific medical knowledge, which can potentially reduce clinical accuracy. Although recent approaches combining imaging models with LLM reasoning have improved reporting, they typically rely on a single generalist agent, restricting their capacity to emulate the diverse and complex reasoning found in multidisciplinary medical teams. To address these limitations, we propose MedChat, a multi-agent diagnostic framework and platform that combines specialized vision models with multiple role-specific LLM agents, all coordinated by a director agent. This design enhances reliability, reduces hallucination risk, and enables interactive diagnostic reporting through an interface tailored for clinical review and educational use. Code available at https://github.com/Purdue-M2/MedChat. | [
{
"version": "v1",
"created": "Mon, 09 Jun 2025 03:51:18 GMT"
},
{
"version": "v2",
"created": "Wed, 11 Jun 2025 04:43:27 GMT"
}
] | 2025-06-12T00:00:00 | [
[
"Liu",
"Philip R.",
""
],
[
"Bansal",
"Sparsh",
""
],
[
"Dinh",
"Jimmy",
""
],
[
"Pawar",
"Aditya",
""
],
[
"Satishkumar",
"Ramani",
""
],
[
"Desai",
"Shail",
""
],
[
"Gupta",
"Neeraj",
""
],
[
"Wang",
"Xin",
""
],
[
"Hu",
"Shu",
""
]
] |
2506.07410 | Dongzhe Zheng | Dongzhe Zheng | Spencer Differential Degeneration Theory and Its Applications in Algebraic Geometry | null | null | null | null | math.AG math-ph math.DG math.MP | http://creativecommons.org/licenses/by/4.0/ | Based on the compatible pair theory of principal bundle constraint systems, this paper discovers and establishes a complete Spencer differential degeneration theory. We prove that when symmetric tensors satisfy a $\lambda$-dependent kernel condition $\delta_{\mathfrak{g}}^{\lambda}(s)=0$, the Spencer differential degenerates to the standard exterior differential, thus establishing a precise bridge between the complex Spencer theory and the classical de Rham theory. One of the advances in this paper is the rigorous proof that this degeneration condition remains stable under mirror transformations, revealing the profound symmetry origins of this phenomenon. Based on these rigorous mathematical results, we construct a canonical mapping from degenerate Spencer cocycles to de Rham cohomology and elucidate its geometric meaning. Finally, we demonstrate the application potential of this theory in algebraic geometry, particularly on K3 surfaces, where we preliminarily verify that this framework can systematically identify (1,1)-Hodge classes satisfying algebraicity conditions. This work provides new perspectives and technical approaches for studying algebraic invariants using tools from constraint geometry. | [
{
"version": "v1",
"created": "Mon, 09 Jun 2025 04:13:57 GMT"
},
{
"version": "v2",
"created": "Wed, 11 Jun 2025 06:43:26 GMT"
}
] | 2025-06-12T00:00:00 | [
[
"Zheng",
"Dongzhe",
""
]
] |
2506.07459 | Jiajun Fan | Ziwen Wang, Jiajun Fan, Ruihan Guo, Thao Nguyen, Heng Ji, Ge Liu | ProteinZero: Self-Improving Protein Generation via Online Reinforcement Learning | null | null | null | null | cs.LG q-bio.QM | http://creativecommons.org/licenses/by/4.0/ | Protein generative models have shown remarkable promise in protein design but still face limitations in success rate, due to the scarcity of high-quality protein datasets for supervised pretraining. We present ProteinZero, a novel framework that enables scalable, automated, and continuous self-improvement of the inverse folding model through online reinforcement learning. To achieve computationally tractable online feedback, we introduce efficient proxy reward models based on ESM-fold and a novel rapid ddG predictor that significantly accelerates evaluation speed. ProteinZero employs a general RL framework balancing multi-reward maximization, KL-divergence from a reference model, and a novel protein-embedding level diversity regularization that prevents mode collapse while promoting higher sequence diversity. Through extensive experiments, we demonstrate that ProteinZero substantially outperforms existing methods across every key metric in protein design, achieving significant improvements in structural accuracy, designability, thermodynamic stability, and sequence diversity. Most impressively, ProteinZero reduces design failure rates by approximately 36% - 48% compared to widely-used methods like ProteinMPNN, ESM-IF and InstructPLM, consistently achieving success rates exceeding 90% across diverse and complex protein folds. Notably, the entire RL run on CATH-4.3 can be done with a single 8 X GPU node in under 3 days, including reward computation. Our work establishes a new paradigm for protein design where models evolve continuously from their own generated outputs, opening new possibilities for exploring the vast protein design space. | [
{
"version": "v1",
"created": "Mon, 09 Jun 2025 06:08:59 GMT"
},
{
"version": "v2",
"created": "Tue, 10 Jun 2025 18:30:51 GMT"
}
] | 2025-06-12T00:00:00 | [
[
"Wang",
"Ziwen",
""
],
[
"Fan",
"Jiajun",
""
],
[
"Guo",
"Ruihan",
""
],
[
"Nguyen",
"Thao",
""
],
[
"Ji",
"Heng",
""
],
[
"Liu",
"Ge",
""
]
] |
2506.07473 | Emmanuel Deruty | Emmanuel Deruty | An introduction to pitch strength in contemporary popular music analysis and production | In Music 2024, Innovation in Music Conference, 14-16 June, 2024, Kristiania University College, Oslo, Norway | null | null | null | cs.SD eess.AS | http://creativecommons.org/licenses/by/4.0/ | Music information retrieval distinguishes between low- and high-level descriptions of music. Current generative AI models rely on text descriptions that are higher level than the controls familiar to studio musicians. Pitch strength, a low-level perceptual parameter of contemporary popular music, may be one feature that could make such AI models more suited to music production. Signal and perceptual analyses suggest that pitch strength (1) varies significantly across and inside songs; (2) contributes to both small- and large-scale structure; (3) contributes to the handling of polyphonic dissonance; and (4) may be a feature of upper harmonics made audible in a perspective of perceptual richness. | [
{
"version": "v1",
"created": "Mon, 09 Jun 2025 06:47:26 GMT"
},
{
"version": "v2",
"created": "Tue, 10 Jun 2025 15:46:29 GMT"
},
{
"version": "v3",
"created": "Wed, 11 Jun 2025 06:10:16 GMT"
}
] | 2025-06-12T00:00:00 | [
[
"Deruty",
"Emmanuel",
""
]
] |
2506.07494 | Najm Hassan | Peng Huang, Imdad Ullah, Xiaotong Wei, Tariq Ahamed Ahanger, Najm Hassan, Zawar Hussain Shah | Towards Energy-Efficient and Low-Latency Voice-Controlled Smart Homes: A Proposal for Offline Speech Recognition and IoT Integration | null | null | null | null | cs.SD cs.CY eess.AS | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | The smart home systems, based on AI speech recognition and IoT technology, enable people to control devices through verbal commands and make people's lives more efficient. However, existing AI speech recognition services are primarily deployed on cloud platforms on the Internet. When users issue a command, speech recognition devices like ``Amazon Echo'' will post a recording through numerous network nodes, reach multiple servers, and then receive responses through the Internet. This mechanism presents several issues, including unnecessary energy consumption, communication latency, and the risk of a single-point failure. In this position paper, we propose a smart home concept based on offline speech recognition and IoT technology: 1) integrating offline keyword spotting (KWS) technologies into household appliances with limited resource hardware to enable them to understand user voice commands; 2) designing a local IoT network with decentralized architecture to manage and connect various devices, enhancing the robustness and scalability of the system. This proposal of a smart home based on offline speech recognition and IoT technology will allow users to use low-latency voice control anywhere in the home without depending on the Internet and provide better scalability and energy sustainability. | [
{
"version": "v1",
"created": "Mon, 09 Jun 2025 07:15:48 GMT"
},
{
"version": "v2",
"created": "Wed, 11 Jun 2025 12:50:56 GMT"
}
] | 2025-06-12T00:00:00 | [
[
"Huang",
"Peng",
""
],
[
"Ullah",
"Imdad",
""
],
[
"Wei",
"Xiaotong",
""
],
[
"Ahanger",
"Tariq Ahamed",
""
],
[
"Hassan",
"Najm",
""
],
[
"Shah",
"Zawar Hussain",
""
]
] |
2506.07497 | Guo Xiangyu | Xiangyu Guo and Zhanqian Wu and Kaixin Xiong and Ziyang Xu and Lijun Zhou and Gangwei Xu and Shaoqing Xu and Haiyang Sun and Bing Wang and Guang Chen and Hangjun Ye and Wenyu Liu and Xinggang Wang | Genesis: Multimodal Driving Scene Generation with Spatio-Temporal and Cross-Modal Consistency | null | null | null | null | cs.CV | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | We present Genesis, a unified framework for joint generation of multi-view driving videos and LiDAR sequences with spatio-temporal and cross-modal consistency. Genesis employs a two-stage architecture that integrates a DiT-based video diffusion model with 3D-VAE encoding, and a BEV-aware LiDAR generator with NeRF-based rendering and adaptive sampling. Both modalities are directly coupled through a shared latent space, enabling coherent evolution across visual and geometric domains. To guide the generation with structured semantics, we introduce DataCrafter, a captioning module built on vision-language models that provides scene-level and instance-level supervision. Extensive experiments on the nuScenes benchmark demonstrate that Genesis achieves state-of-the-art performance across video and LiDAR metrics (FVD 16.95, FID 4.24, Chamfer 0.611), and benefits downstream tasks including segmentation and 3D detection, validating the semantic fidelity and practical utility of the generated data. | [
{
"version": "v1",
"created": "Mon, 09 Jun 2025 07:20:49 GMT"
},
{
"version": "v2",
"created": "Wed, 11 Jun 2025 08:38:38 GMT"
}
] | 2025-06-12T00:00:00 | [
[
"Guo",
"Xiangyu",
""
],
[
"Wu",
"Zhanqian",
""
],
[
"Xiong",
"Kaixin",
""
],
[
"Xu",
"Ziyang",
""
],
[
"Zhou",
"Lijun",
""
],
[
"Xu",
"Gangwei",
""
],
[
"Xu",
"Shaoqing",
""
],
[
"Sun",
"Haiyang",
""
],
[
"Wang",
"Bing",
""
],
[
"Chen",
"Guang",
""
],
[
"Ye",
"Hangjun",
""
],
[
"Liu",
"Wenyu",
""
],
[
"Wang",
"Xinggang",
""
]
] |
2506.07547 | Miura Chiaki | Chiaki Miura, Ichiro Sakata | From Rapid Release to Reinforced Elite: Citation Inequality Is Stronger in Preprints than Journals | null | null | null | null | cs.DL | http://creativecommons.org/licenses/by/4.0/ | Preprints have been considered primarily as a supplement to journal-based systems for the rapid dissemination of relevant scientific knowledge and have historically been supported by studies indicating that preprints and published reports have comparable authorship, references, and quality. However, as preprints increasingly serve as an independent medium for scholarly communication rather than precursors to the version of record, it remains uncertain how preprint usage is shaping scientific discourse. Our research revealed that the preprint citations exhibit significantly higher inequality than journal citations, consistently among categories. This trend persisted even when controlling for age and the mean citation count of the journal matched to each of the preprint categories. We also found that the citation inequality in preprints is not solely driven by a few highly cited papers or those with no impact, but rather reflects a broader systemic effect. Whether the preprint is subsequently published in a journal or not does not significantly affect the citation inequality. Further analyses of the structural factors show that preferential attachment does not significantly contribute to citation inequality in preprints, whereas author prestige plays a substantial role. Notably, the gap in citation inequality between the preprint category and the journal is more pronounced in fields where preprints are more established, such as mathematics, physics, and high-energy physics. This highlights a potential vulnerability in preprint ecosystems where reputation-driven citation may hinder scientific diversity. | [
{
"version": "v1",
"created": "Mon, 09 Jun 2025 08:38:17 GMT"
},
{
"version": "v2",
"created": "Wed, 11 Jun 2025 12:47:21 GMT"
}
] | 2025-06-12T00:00:00 | [
[
"Miura",
"Chiaki",
""
],
[
"Sakata",
"Ichiro",
""
]
] |
2506.07563 | Eyal German | Ken Yaggel, Eyal German and Aviel Ben Siman Tov | MoE-MLoRA for Multi-Domain CTR Prediction: Efficient Adaptation with Expert Specialization | null | null | null | null | cs.IR cs.AI | http://creativecommons.org/licenses/by/4.0/ | Personalized recommendation systems must adapt to user interactions across different domains. Traditional approaches like MLoRA apply a single adaptation per domain but lack flexibility in handling diverse user behaviors. To address this, we propose MoE-MLoRA, a mixture-of-experts framework where each expert is first trained independently to specialize in its domain before a gating network is trained to weight their contributions dynamically. We evaluate MoE-MLoRA across eight CTR models on Movielens and Taobao, showing that it improves performance in large-scale, dynamic datasets (+1.45 Weighed-AUC in Taobao-20) but offers limited benefits in structured datasets with low domain diversity and sparsity. Further analysis of the number of experts per domain reveals that larger ensembles do not always improve performance, indicating the need for model-aware tuning. Our findings highlight the potential of expert-based architectures for multi-domain recommendation systems, demonstrating that task-aware specialization and adaptive gating can enhance predictive accuracy in complex environments. The implementation and code are available in our GitHub repository. | [
{
"version": "v1",
"created": "Mon, 09 Jun 2025 09:03:05 GMT"
},
{
"version": "v2",
"created": "Tue, 10 Jun 2025 06:56:22 GMT"
},
{
"version": "v3",
"created": "Wed, 11 Jun 2025 07:55:32 GMT"
}
] | 2025-06-12T00:00:00 | [
[
"Yaggel",
"Ken",
""
],
[
"German",
"Eyal",
""
],
[
"Tov",
"Aviel Ben Siman",
""
]
] |
2506.07564 | Xinkai Zou | Peiran Li, Xinkai Zou, Zhuohang Wu, Ruifeng Li, Shuo Xing, Hanwen Zheng, Zhikai Hu, Yuping Wang, Haoxi Li, Qin Yuan, Yingmo Zhang, Zhengzhong Tu | SAFEFLOW: A Principled Protocol for Trustworthy and Transactional Autonomous Agent Systems | Former versions either contain unrelated content or cannot be properly converted to PDF | null | null | null | cs.AI cs.CL | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Recent advances in large language models (LLMs) and vision-language models (VLMs) have enabled powerful autonomous agents capable of complex reasoning and multi-modal tool use. Despite their growing capabilities, today's agent frameworks remain fragile, lacking principled mechanisms for secure information flow, reliability, and multi-agent coordination. In this work, we introduce SAFEFLOW, a new protocol-level framework for building trustworthy LLM/VLM-based agents. SAFEFLOW enforces fine-grained information flow control (IFC), precisely tracking provenance, integrity, and confidentiality of all the data exchanged between agents, tools, users, and environments. By constraining LLM reasoning to respect these security labels, SAFEFLOW prevents untrusted or adversarial inputs from contaminating high-integrity decisions. To ensure robustness in concurrent multi-agent settings, SAFEFLOW introduces transactional execution, conflict resolution, and secure scheduling over shared state, preserving global consistency across agents. We further introduce mechanisms, including write-ahead logging, rollback, and secure caches, that further enhance resilience against runtime errors and policy violations. To validate the performances, we built SAFEFLOWBENCH, a comprehensive benchmark suite designed to evaluate agent reliability under adversarial, noisy, and concurrent operational conditions. Extensive experiments demonstrate that agents built with SAFEFLOW maintain impressive task performance and security guarantees even in hostile environments, substantially outperforming state-of-the-art. Together, SAFEFLOW and SAFEFLOWBENCH lay the groundwork for principled, robust, and secure agent ecosystems, advancing the frontier of reliable autonomy. | [
{
"version": "v1",
"created": "Mon, 09 Jun 2025 09:04:37 GMT"
},
{
"version": "v2",
"created": "Tue, 10 Jun 2025 13:47:37 GMT"
},
{
"version": "v3",
"created": "Wed, 11 Jun 2025 03:14:10 GMT"
}
] | 2025-06-12T00:00:00 | [
[
"Li",
"Peiran",
""
],
[
"Zou",
"Xinkai",
""
],
[
"Wu",
"Zhuohang",
""
],
[
"Li",
"Ruifeng",
""
],
[
"Xing",
"Shuo",
""
],
[
"Zheng",
"Hanwen",
""
],
[
"Hu",
"Zhikai",
""
],
[
"Wang",
"Yuping",
""
],
[
"Li",
"Haoxi",
""
],
[
"Yuan",
"Qin",
""
],
[
"Zhang",
"Yingmo",
""
],
[
"Tu",
"Zhengzhong",
""
]
] |
2506.07584 | Hao Li | Hao Li, Bowen Deng, Chang Xu, Zhiyuan Feng, Viktor Schlegel, Yu-Hao Huang, Yizheng Sun, Jingyuan Sun, Kailai Yang, Yiyao Yu, Jiang Bian | MIRA: Medical Time Series Foundation Model for Real-World Health Data | null | null | null | null | cs.LG | http://creativecommons.org/licenses/by-nc-sa/4.0/ | A unified foundation model for medical time series -- pretrained on open access and ethics board-approved medical corpora -- offers the potential to reduce annotation burdens, minimize model customization, and enable robust transfer across clinical institutions, modalities, and tasks, particularly in data-scarce or privacy-constrained environments. However, existing generalist time series foundation models struggle to handle medical time series data due to their inherent challenges, including irregular intervals, heterogeneous sampling rates, and frequent missing values. To address these challenges, we introduce MIRA, a unified foundation model specifically designed for medical time series forecasting. MIRA incorporates a Continuous-Time Rotary Positional Encoding that enables fine-grained modeling of variable time intervals, a frequency-specific mixture-of-experts layer that routes computation across latent frequency regimes to further promote temporal specialization, and a Continuous Dynamics Extrapolation Block based on Neural ODE that models the continuous trajectory of latent states, enabling accurate forecasting at arbitrary target timestamps. Pretrained on a large-scale and diverse medical corpus comprising over 454 billion time points collect from publicly available datasets, MIRA achieves reductions in forecasting errors by an average of 10% and 7% in out-of-distribution and in-distribution scenarios, respectively, when compared to other zero-shot and fine-tuned baselines. We also introduce a comprehensive benchmark spanning multiple downstream clinical tasks, establishing a foundation for future research in medical time series modeling. | [
{
"version": "v1",
"created": "Mon, 09 Jun 2025 09:27:17 GMT"
},
{
"version": "v2",
"created": "Wed, 11 Jun 2025 08:03:59 GMT"
}
] | 2025-06-12T00:00:00 | [
[
"Li",
"Hao",
""
],
[
"Deng",
"Bowen",
""
],
[
"Xu",
"Chang",
""
],
[
"Feng",
"Zhiyuan",
""
],
[
"Schlegel",
"Viktor",
""
],
[
"Huang",
"Yu-Hao",
""
],
[
"Sun",
"Yizheng",
""
],
[
"Sun",
"Jingyuan",
""
],
[
"Yang",
"Kailai",
""
],
[
"Yu",
"Yiyao",
""
],
[
"Bian",
"Jiang",
""
]
] |
2506.07664 | Lei Xu | Lei Xu, Sirui Chen, Yuxuan Huang, Chaochao Lu | Synthesis by Design: Controlled Data Generation via Structural Guidance | null | null | null | null | cs.CL cs.AI | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Mathematical reasoning remains challenging for LLMs due to complex logic and the need for precise computation. Existing methods enhance LLM reasoning by synthesizing datasets through problem rephrasing, but face issues with generation quality and problem complexity. To address this, we propose to extract structural information with generated problem-solving code from mathematical reasoning and guide data generation with structured solutions. Applied to MATH and GSM8K, our approach produces 39K problems with labeled intermediate steps and a 6.1K-problem benchmark of higher difficulty. Results on our benchmark show that model performance declines as reasoning length increases. Additionally, we conducted fine-tuning experiments using the proposed training data on a range of LLMs, and the results validate the effectiveness of our dataset. We hope the proposed method and dataset will contribute to future research in enhancing LLM reasoning capabilities. Our code and data are available at https://github.com/OpenCausaLab/StructuralGeneration. | [
{
"version": "v1",
"created": "Mon, 09 Jun 2025 11:38:23 GMT"
},
{
"version": "v2",
"created": "Tue, 10 Jun 2025 21:19:15 GMT"
}
] | 2025-06-12T00:00:00 | [
[
"Xu",
"Lei",
""
],
[
"Chen",
"Sirui",
""
],
[
"Huang",
"Yuxuan",
""
],
[
"Lu",
"Chaochao",
""
]
] |
2506.07711 | Guillaume Maitrier | Guillaume Maitrier, Jean-Philippe Bouchaud | The Subtle Interplay between Square-root Impact, Order Imbalance & Volatility: A Unifying Framework | null | null | null | null | q-fin.TR q-fin.ST | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | In this work, we aim to reconcile several apparently contradictory observations in market microstructure: is the famous ''square-root law'' of metaorder impact that decays with time compatible with the random-walk nature of prices and the linear impact of order imbalances? Can one entirely explain the volatility of prices as resulting from the flow of uninformed metaorders that mechanically impact prices? We introduce a new theoretical framework to describe metaorders with different signs, sizes and durations, which all impact prices as a square-root of volume but with a subsequent time decay. We show that, as in the original propagator model, price diffusion is ensured by the long memory of cross-correlations between metaorders. In order to account for the effect of strongly fluctuating volumes $q$ of individual trades, we further introduce two $q$-dependent exponents, which allows us to account for the way the moments of generalized volume imbalance and the correlation between price changes and generalized order flow imbalance scales with $T$. We predict in particular that the corresponding power-laws depend in a non-monotonic fashion on a parameter $a$ that allows one to put the same weight on all child orders or overweight large orders, a behaviour clearly borne out by empirical data. We also predict that the correlation between price changes and volume imbalances should display a maximum as a function of $a$, which again matches observations. Such noteworthy agreement between theory and data suggests that our framework correctly captures the basic mechanism at the heart of price formation, namely the average impact of metaorders. We argue that our results support the ''Order-Driven'' theory of excess volatility, and are at odds with the idea that a ''Fundamental'' component accounts for a large share of the volatility of financial markets. | [
{
"version": "v1",
"created": "Mon, 09 Jun 2025 12:53:25 GMT"
},
{
"version": "v2",
"created": "Tue, 10 Jun 2025 14:15:17 GMT"
},
{
"version": "v3",
"created": "Wed, 11 Jun 2025 07:04:05 GMT"
}
] | 2025-06-12T00:00:00 | [
[
"Maitrier",
"Guillaume",
""
],
[
"Bouchaud",
"Jean-Philippe",
""
]
] |
2506.07721 | Kazuki Oi | Kazuki Oi, Shimpei Endo | Universal Efimov spectra and fermionic doublets in highly mass-imbalanced cold-atom mixtures with van der Waals and dipole interactions | 10 pages, 4 figures | null | null | null | cond-mat.quant-gas nucl-th | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | We study the Efimov states in highly mass-imbalanced three-body systems composed of two identical heavy atoms and one light atom, focusing on the Er-Er-Li and Dy-Dy-Li cold-atom mixtures with strong dipole-dipole interactions between the heavy atoms. By solving the Born-Oppenheimer equation for varying $s$-wave scattering lengths between the heavy and light atoms, we demonstrate for both bosonic and fermionic systems that the Efimov spectra and hence the three-body parameters are universal even with the dipole interaction comparable in strength to the van der Waals interaction. While the bosonic systems exhibit Efimov states only in the $M_z=0$ channel, the fermionic systems show a characteristic doublet of the Efimov states in the $M_z=0$ and $M_z = \pm 1$ channels due to the interplay of finite angular momentum and the anisotropy of the dipole interaction. Both numerical results and analytical formula obtained with the first-order perturbation show that the ratio of the three-body parameters between these two fermionic channels exhibits universality, particularly well in the limit of large mass imbalance. Leveraging this universality, we provide quantitative predictions for the values and ratios of the three-body parameters for experimentally relevant Er-Li and Dy-Li isotopes. | [
{
"version": "v1",
"created": "Mon, 09 Jun 2025 13:03:57 GMT"
}
] | 2025-06-12T00:00:00 | [
[
"Oi",
"Kazuki",
""
],
[
"Endo",
"Shimpei",
""
]
] |
2506.07736 | Jingnan Zheng | Jingnan Zheng, Xiangtian Ji, Yijun Lu, Chenhang Cui, Weixiang Zhao, Gelei Deng, Zhenkai Liang, An Zhang, Tat-Seng Chua | RSafe: Incentivizing proactive reasoning to build robust and adaptive LLM safeguards | null | null | null | null | cs.AI | http://creativecommons.org/licenses/by/4.0/ | Large Language Models (LLMs) continue to exhibit vulnerabilities despite deliberate safety alignment efforts, posing significant risks to users and society. To safeguard against the risk of policy-violating content, system-level moderation via external guard models-designed to monitor LLM inputs and outputs and block potentially harmful content-has emerged as a prevalent mitigation strategy. Existing approaches of training guard models rely heavily on extensive human curated datasets and struggle with out-of-distribution threats, such as emerging harmful categories or jailbreak attacks. To address these limitations, we propose RSafe, an adaptive reasoning-based safeguard that conducts guided safety reasoning to provide robust protection within the scope of specified safety policies. RSafe operates in two stages: 1) guided reasoning, where it analyzes safety risks of input content through policy-guided step-by-step reasoning, and 2) reinforced alignment, where rule-based RL optimizes its reasoning paths to align with accurate safety prediction. This two-stage training paradigm enables RSafe to internalize safety principles to generalize safety protection capability over unseen or adversarial safety violation scenarios. During inference, RSafe accepts user-specified safety policies to provide enhanced safeguards tailored to specific safety requirements. | [
{
"version": "v1",
"created": "Mon, 09 Jun 2025 13:20:04 GMT"
},
{
"version": "v2",
"created": "Wed, 11 Jun 2025 09:27:11 GMT"
}
] | 2025-06-12T00:00:00 | [
[
"Zheng",
"Jingnan",
""
],
[
"Ji",
"Xiangtian",
""
],
[
"Lu",
"Yijun",
""
],
[
"Cui",
"Chenhang",
""
],
[
"Zhao",
"Weixiang",
""
],
[
"Deng",
"Gelei",
""
],
[
"Liang",
"Zhenkai",
""
],
[
"Zhang",
"An",
""
],
[
"Chua",
"Tat-Seng",
""
]
] |
2506.07737 | Huamin Wang | Xuemei Chen, Huamin Wang, Hangchi Shen, Shukai Duan, Shiping Wen, Tingwen Huang | SpikeSMOKE: Spiking Neural Networks for Monocular 3D Object Detection with Cross-Scale Gated Coding | null | null | null | null | cs.CV | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Low energy consumption for 3D object detection is an important research area because of the increasing energy consumption with their wide application in fields such as autonomous driving. The spiking neural networks (SNNs) with low-power consumption characteristics can provide a novel solution for this research. Therefore, we apply SNNs to monocular 3D object detection and propose the SpikeSMOKE architecture in this paper, which is a new attempt for low-power monocular 3D object detection. As we all know, discrete signals of SNNs will generate information loss and limit their feature expression ability compared with the artificial neural networks (ANNs).In order to address this issue, inspired by the filtering mechanism of biological neuronal synapses, we propose a cross-scale gated coding mechanism(CSGC), which can enhance feature representation by combining cross-scale fusion of attentional methods and gated filtering mechanisms.In addition, to reduce the computation and increase the speed of training, we present a novel light-weight residual block that can maintain spiking computing paradigm and the highest possible detection performance. Compared to the baseline SpikeSMOKE under the 3D Object Detection, the proposed SpikeSMOKE with CSGC can achieve 11.78 (+2.82, Easy), 10.69 (+3.2, Moderate), and 10.48 (+3.17, Hard) on the KITTI autonomous driving dataset by AP|R11 at 0.7 IoU threshold, respectively. It is important to note that the results of SpikeSMOKE can significantly reduce energy consumption compared to the results on SMOKE. For example,the energy consumption can be reduced by 72.2% on the hard category, while the detection performance is reduced by only 4%. SpikeSMOKE-L (lightweight) can further reduce the amount of parameters by 3 times and computation by 10 times compared to SMOKE. | [
{
"version": "v1",
"created": "Mon, 09 Jun 2025 13:20:10 GMT"
},
{
"version": "v2",
"created": "Wed, 11 Jun 2025 15:55:43 GMT"
}
] | 2025-06-12T00:00:00 | [
[
"Chen",
"Xuemei",
""
],
[
"Wang",
"Huamin",
""
],
[
"Shen",
"Hangchi",
""
],
[
"Duan",
"Shukai",
""
],
[
"Wen",
"Shiping",
""
],
[
"Huang",
"Tingwen",
""
]
] |
2506.07745 | Abdul Sattar | Sana Maroof, Abdul Sattar, Azmat Iqbal Bashir, Muhammad Irfan, Hamid Latif, Hina Mustafa, Ahmad Saeed, Raja Junaid Amjad, Farah Alvi | First-principles Quantum Insights into Bandgap Engineering, Valley Quantum Hall Effect, and Nonlinear Optical Response of Ge-Doped Graphene for Potential Optoelectronic Applications | 21 pages, 10 figure | null | null | null | physics.comp-ph | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | The valley in the band structure of materials has gained a lot of attention recently. The promising applications of the valley degree of freedom include the next-generation valleytronic devices, quantum information processing, quantum computing, and optoelectronic devices. Graphene is an ideal quantum material for high-speed valleytronic applications because of its high carrier mobility and convenience of bandgap engineering. Employing first-principles density functional theoretical approach, this study opted bandgap engineering strategy via Germanium doping to open bandgap and enhance valley selectivity in graphene monolayers. The impact of Ge dopant concentration of 2%, 3.125%, 5.5%, and 12.5% is explored on the valleytronic; valley Hall effect, valley transport, and optical properties. The reported results demonstrate that bandgap, valley polarization, and second harmonic generation can be tuned effectively by varying doping concentration of Germanium. The Berry curvature profile is antisymmetric for corresponding K and K' valleys, thus leading to valley-dependent transport properties and a potential valley Hall effect. Finally, the second-order susceptibilities exhibit corresponding optical absorption peaks, indicating efficient second-harmonic generation due to the broken inversion symmetry. These findings highlight the potential of Ge-doped graphene for nonlinear optics and valleytronics applications, while providing novel insights into its topological phase and transport properties. | [
{
"version": "v1",
"created": "Mon, 09 Jun 2025 13:26:25 GMT"
},
{
"version": "v2",
"created": "Wed, 11 Jun 2025 10:37:41 GMT"
}
] | 2025-06-12T00:00:00 | [
[
"Maroof",
"Sana",
""
],
[
"Sattar",
"Abdul",
""
],
[
"Bashir",
"Azmat Iqbal",
""
],
[
"Irfan",
"Muhammad",
""
],
[
"Latif",
"Hamid",
""
],
[
"Mustafa",
"Hina",
""
],
[
"Saeed",
"Ahmad",
""
],
[
"Amjad",
"Raja Junaid",
""
],
[
"Alvi",
"Farah",
""
]
] |
2506.07751 | Silin Gao | Silin Gao, Antoine Bosselut, Samy Bengio, Emmanuel Abbe | AbstRaL: Augmenting LLMs' Reasoning by Reinforcing Abstract Thinking | Under review | null | null | null | cs.CL cs.AI cs.SC | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Recent studies have shown that large language models (LLMs), especially smaller ones, often lack robustness in their reasoning. I.e., they tend to experience performance drops when faced with distribution shifts, such as changes to numerical or nominal variables, or insertions of distracting clauses. A possible strategy to address this involves generating synthetic data to further "instantiate" reasoning problems on potential variations. In contrast, our approach focuses on "abstracting" reasoning problems. This not only helps counteract distribution shifts but also facilitates the connection to symbolic tools for deriving solutions. We find that this abstraction process is better acquired through reinforcement learning (RL) than just supervised fine-tuning, which often fails to produce faithful abstractions. Our method, AbstRaL -- which promotes abstract reasoning in LLMs using RL on granular abstraction data -- significantly mitigates performance degradation on recent GSM perturbation benchmarks. | [
{
"version": "v1",
"created": "Mon, 09 Jun 2025 13:34:50 GMT"
},
{
"version": "v2",
"created": "Wed, 11 Jun 2025 00:45:46 GMT"
}
] | 2025-06-12T00:00:00 | [
[
"Gao",
"Silin",
""
],
[
"Bosselut",
"Antoine",
""
],
[
"Bengio",
"Samy",
""
],
[
"Abbe",
"Emmanuel",
""
]
] |
2506.07798 | Andrew Hughes | Andrew K. Hughes, Francesco Carotenuto, Thomas D. Russell, Alexandra J. Tetarenko, James C. A. Miller-Jones, Arash Bahramian, Joe S. Bright, Fraser J. Cowie, Rob Fender, Mark A. Gurwell, Jasvinderjit K. Khaulsay, Anastasia Kirby, Serena Jones, Elodie Lescure, Michael McCollough, Richard M. Plotkin, Ramprasad Rao, Saeqa D. Vrtilek, David R. A. Williams-Baldwin, Callan M. Wood, Gregory R. Sivakoff, Diego Altamirano, Piergiorgio Casella, Stephane Corbel, David R. DeBoer, Melania Del Santo, Constanza Echiburu-Trujillo, Wael Farah, Poshak Gandhi, Karri I. I. Koljonen, Thomas Maccarone, James H. Matthews, Sera B. Markoff, Alexander W. Pollak, David M. Russell, Payaswini Saikia, Noel Castro Segura, Aarran W. Shaw, Andrew Siemion, Roberto Soria, John A. Tomsick, and Jakob van den Eijnden | Comprehensive Radio Monitoring of the Black Hole X-ray Binary Swift J1727.8$-$1613 during its 2023$-$2024 Outburst | null | null | 10.3847/1538-4357/ade2e6 | null | astro-ph.HE | http://creativecommons.org/licenses/by/4.0/ | This work presents comprehensive multi-frequency radio monitoring of the black hole low-mass X-ray binary Swift J1727.8$-$1613, which underwent its first recorded outburst after its discovery in August 2023. Through a considerable community effort, we have coalesced the data from multiple, distinct observing programs; the light curves include ${\sim} 10$ months and 197 epochs of monitoring from 7 radio facilities with observing frequencies ranging from (approximately) 0.3$-$230GHz. The primary purpose of this work is to provide the broader astronomical community with these light curves to assist with the interpretation of other observing campaigns, particularly non-radio observing frequencies. We discuss the phenomenological evolution of the source, which included: (i) multiple radio flares consistent with the launching of discrete jet ejections, the brightest of which reached $\sim$ 1 Jy; (ii) temporally evolving radio spectral indices ($\alpha$), reaching values steeper than expected for optically-thin synchrotron emission ($\alpha {<} -1$) and emission with significant radiative cooling ($\alpha < -1.5$). We have published a digital copy of the data and intend for this work to set a precedent for the community to continue releasing comprehensive radio light curves of future low-mass X-ray binary outbursts. | [
{
"version": "v1",
"created": "Mon, 09 Jun 2025 14:26:00 GMT"
}
] | 2025-06-12T00:00:00 | [
[
"Hughes",
"Andrew K.",
""
],
[
"Carotenuto",
"Francesco",
""
],
[
"Russell",
"Thomas D.",
""
],
[
"Tetarenko",
"Alexandra J.",
""
],
[
"Miller-Jones",
"James C. A.",
""
],
[
"Bahramian",
"Arash",
""
],
[
"Bright",
"Joe S.",
""
],
[
"Cowie",
"Fraser J.",
""
],
[
"Fender",
"Rob",
""
],
[
"Gurwell",
"Mark A.",
""
],
[
"Khaulsay",
"Jasvinderjit K.",
""
],
[
"Kirby",
"Anastasia",
""
],
[
"Jones",
"Serena",
""
],
[
"Lescure",
"Elodie",
""
],
[
"McCollough",
"Michael",
""
],
[
"Plotkin",
"Richard M.",
""
],
[
"Rao",
"Ramprasad",
""
],
[
"Vrtilek",
"Saeqa D.",
""
],
[
"Williams-Baldwin",
"David R. A.",
""
],
[
"Wood",
"Callan M.",
""
],
[
"Sivakoff",
"Gregory R.",
""
],
[
"Altamirano",
"Diego",
""
],
[
"Casella",
"Piergiorgio",
""
],
[
"Corbel",
"Stephane",
""
],
[
"DeBoer",
"David R.",
""
],
[
"Del Santo",
"Melania",
""
],
[
"Echiburu-Trujillo",
"Constanza",
""
],
[
"Farah",
"Wael",
""
],
[
"Gandhi",
"Poshak",
""
],
[
"Koljonen",
"Karri I. I.",
""
],
[
"Maccarone",
"Thomas",
""
],
[
"Matthews",
"James H.",
""
],
[
"Markoff",
"Sera B.",
""
],
[
"Pollak",
"Alexander W.",
""
],
[
"Russell",
"David M.",
""
],
[
"Saikia",
"Payaswini",
""
],
[
"Segura",
"Noel Castro",
""
],
[
"Shaw",
"Aarran W.",
""
],
[
"Siemion",
"Andrew",
""
],
[
"Soria",
"Roberto",
""
],
[
"Tomsick",
"John A.",
""
],
[
"Eijnden",
"Jakob van den",
""
]
] |
2506.07866 | Narendra N. Hegade | Sebasti\'an V. Romero, Alejandro Gomez Cadavid, Pavle Nika\v{c}evi\'c, Enrique Solano, Narendra N. Hegade, Miguel Angel Lopez-Ruiz, Claudio Girotto, Masako Yamada, Panagiotis Kl. Barkoutsos, Ananth Kaushik, Martin Roetteler | Protein folding with an all-to-all trapped-ion quantum computer | 12 pages, 4 figures, 2 tables | null | null | null | quant-ph | http://creativecommons.org/licenses/by/4.0/ | We experimentally demonstrate that the bias-field digitized counterdiabatic quantum optimization (BF-DCQO) algorithm, implemented on IonQ's fully connected trapped-ion quantum processors, offers an efficient approach to solving dense higher-order unconstrained binary optimization (HUBO) problems. Specifically, we tackle protein folding on a tetrahedral lattice for up to 12 amino acids, representing the largest quantum hardware implementations of protein folding problems reported to date. Additionally, we address MAX 4-SAT instances at the computational phase transition and fully connected spin-glass problems using all 36 available qubits. Across all considered cases, our method consistently achieves optimal solutions, highlighting the powerful synergy between non-variational quantum optimization approaches and the intrinsic all-to-all connectivity of trapped-ion architectures. Given the expected scalability of trapped-ion quantum systems, BF-DCQO represents a promising pathway toward practical quantum advantage for dense HUBO problems with significant industrial and scientific relevance. | [
{
"version": "v1",
"created": "Mon, 09 Jun 2025 15:34:02 GMT"
},
{
"version": "v2",
"created": "Tue, 10 Jun 2025 22:46:38 GMT"
}
] | 2025-06-12T00:00:00 | [
[
"Romero",
"Sebastián V.",
""
],
[
"Cadavid",
"Alejandro Gomez",
""
],
[
"Nikačević",
"Pavle",
""
],
[
"Solano",
"Enrique",
""
],
[
"Hegade",
"Narendra N.",
""
],
[
"Lopez-Ruiz",
"Miguel Angel",
""
],
[
"Girotto",
"Claudio",
""
],
[
"Yamada",
"Masako",
""
],
[
"Barkoutsos",
"Panagiotis Kl.",
""
],
[
"Kaushik",
"Ananth",
""
],
[
"Roetteler",
"Martin",
""
]
] |
2506.07876 | Yuxin Chen | Xinghao Zhu, Yuxin Chen, Lingfeng Sun, Farzad Niroui, Simon Le Cleac'h, Jiuguang Wang, Kuan Fang | Versatile Loco-Manipulation through Flexible Interlimb Coordination | null | null | null | null | cs.RO cs.SY eess.SY | http://creativecommons.org/licenses/by/4.0/ | The ability to flexibly leverage limbs for loco-manipulation is essential for enabling autonomous robots to operate in unstructured environments. Yet, prior work on loco-manipulation is often constrained to specific tasks or predetermined limb configurations. In this work, we present Reinforcement Learning for Interlimb Coordination (ReLIC), an approach that enables versatile loco-manipulation through flexible interlimb coordination. The key to our approach is an adaptive controller that seamlessly bridges the execution of manipulation motions and the generation of stable gaits based on task demands. Through the interplay between two controller modules, ReLIC dynamically assigns each limb for manipulation or locomotion and robustly coordinates them to achieve the task success. Using efficient reinforcement learning in simulation, ReLIC learns to perform stable gaits in accordance with the manipulation goals in the real world. To solve diverse and complex tasks, we further propose to interface the learned controller with different types of task specifications, including target trajectories, contact points, and natural language instructions. Evaluated on 12 real-world tasks that require diverse and complex coordination patterns, ReLIC demonstrates its versatility and robustness by achieving a success rate of 78.9% on average. Videos and code can be found at https://relic-locoman.rai-inst.com. | [
{
"version": "v1",
"created": "Mon, 09 Jun 2025 15:49:29 GMT"
},
{
"version": "v2",
"created": "Wed, 11 Jun 2025 01:27:55 GMT"
}
] | 2025-06-12T00:00:00 | [
[
"Zhu",
"Xinghao",
""
],
[
"Chen",
"Yuxin",
""
],
[
"Sun",
"Lingfeng",
""
],
[
"Niroui",
"Farzad",
""
],
[
"Cleac'h",
"Simon Le",
""
],
[
"Wang",
"Jiuguang",
""
],
[
"Fang",
"Kuan",
""
]
] |
2506.07943 | Yizhen Li | Yizhen Li, Dell Zhang, Xuelong Li, Yiqing Shen | Decoupling the Image Perception and Multimodal Reasoning for Reasoning Segmentation with Digital Twin Representations | This work was submitted without the consent of all co-authors. We request withdrawal until all parties agree | null | null | null | cs.CV cs.AI | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Reasoning Segmentation (RS) is a multimodal vision-text task that requires segmenting objects based on implicit text queries, demanding both precise visual perception and vision-text reasoning capabilities. Current RS approaches rely on fine-tuning vision-language models (VLMs) for both perception and reasoning, but their tokenization of images fundamentally disrupts continuous spatial relationships between objects. We introduce DTwinSeger, a novel RS approach that leverages Digital Twin (DT) representation as an intermediate layer to decouple perception from reasoning. Innovatively, DTwinSeger reformulates RS as a two-stage process, where the first transforms the image into a structured DT representation that preserves spatial relationships and semantic properties and then employs a Large Language Model (LLM) to perform explicit reasoning over this representation to identify target objects. We propose a supervised fine-tuning method specifically for LLM with DT representation, together with a corresponding fine-tuning dataset Seg-DT, to enhance the LLM's reasoning capabilities with DT representations. Experiments show that our method can achieve state-of-the-art performance on two image RS benchmarks and three image referring segmentation benchmarks. It yields that DT representation functions as an effective bridge between vision and text, enabling complex multimodal reasoning tasks to be accomplished solely with an LLM. | [
{
"version": "v1",
"created": "Mon, 09 Jun 2025 17:05:02 GMT"
},
{
"version": "v2",
"created": "Wed, 11 Jun 2025 13:48:23 GMT"
}
] | 2025-06-12T00:00:00 | [
[
"Li",
"Yizhen",
""
],
[
"Zhang",
"Dell",
""
],
[
"Li",
"Xuelong",
""
],
[
"Shen",
"Yiqing",
""
]
] |
2506.07986 | Tianlin Pan | Zhengyao Lv, Tianlin Pan, Chenyang Si, Zhaoxi Chen, Wangmeng Zuo, Ziwei Liu, Kwan-Yee K. Wong | Rethinking Cross-Modal Interaction in Multimodal Diffusion Transformers | Project Page: https://vchitect.github.io/TACA/ | null | null | null | cs.CV | http://creativecommons.org/licenses/by/4.0/ | Multimodal Diffusion Transformers (MM-DiTs) have achieved remarkable progress in text-driven visual generation. However, even state-of-the-art MM-DiT models like FLUX struggle with achieving precise alignment between text prompts and generated content. We identify two key issues in the attention mechanism of MM-DiT, namely 1) the suppression of cross-modal attention due to token imbalance between visual and textual modalities and 2) the lack of timestep-aware attention weighting, which hinder the alignment. To address these issues, we propose \textbf{Temperature-Adjusted Cross-modal Attention (TACA)}, a parameter-efficient method that dynamically rebalances multimodal interactions through temperature scaling and timestep-dependent adjustment. When combined with LoRA fine-tuning, TACA significantly enhances text-image alignment on the T2I-CompBench benchmark with minimal computational overhead. We tested TACA on state-of-the-art models like FLUX and SD3.5, demonstrating its ability to improve image-text alignment in terms of object appearance, attribute binding, and spatial relationships. Our findings highlight the importance of balancing cross-modal attention in improving semantic fidelity in text-to-image diffusion models. Our codes are publicly available at \href{https://github.com/Vchitect/TACA} | [
{
"version": "v1",
"created": "Mon, 09 Jun 2025 17:54:04 GMT"
},
{
"version": "v2",
"created": "Wed, 11 Jun 2025 04:30:58 GMT"
}
] | 2025-06-12T00:00:00 | [
[
"Lv",
"Zhengyao",
""
],
[
"Pan",
"Tianlin",
""
],
[
"Si",
"Chenyang",
""
],
[
"Chen",
"Zhaoxi",
""
],
[
"Zuo",
"Wangmeng",
""
],
[
"Liu",
"Ziwei",
""
],
[
"Wong",
"Kwan-Yee K.",
""
]
] |
2506.08000 | Fabian Pichler | Fabian Pichler, Clemens Kuhlenkamp, Michael Knap, Ashvin Vishwanath | Microscopic Mechanism of Anyon Superconductivity Emerging from Fractional Chern Insulators | 19 pages main text + 7 pages Appendix and References; 7 + 2 figures | null | null | null | cond-mat.str-el cond-mat.mes-hall cond-mat.supr-con quant-ph | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Fractional quantum Hall (FQH) states and superconductors typically require contrasting conditions, yet recent experiments have observed them in the same device. A natural explanation is that mobile anyons give rise to superconductivity; however, this mechanism requires binding of minimally charged anyons to establish an unusual energy hierarchy. This scenario has mostly been studied with effective theories, leaving open the question of how anyon superconductivity can arise from repulsive interactions. Here, we show that such an energy hierarchy of anyons arises naturally in fractional Chern insulators (FCIs) at fillings $\nu = 2/(4p \mp 1)$ when they are driven toward a quantum phase transition into a ``semion crystal'' -- an exotic charge-density-wave (CDW) insulator with semion topological order. Near the transition, Cooper-pair correlations are enhanced, so that a conventional charge-2e superconductor appears with doping. Guided by these insights, we analyze a microscopic realization in a repulsive Hubbard-Hofstadter model. Tensor network simulations at $\nu = 2/3$ reveal a robust FCI that, with increasing interactions, transitions into the semion crystal. Finding a stable semion crystal in such a minimal model highlights it as a viable state competing with conventional CDW and FQH states. In the vicinity of this transition, we find markedly enhanced Cooper pairing, consistent with our theory that the 2e/3 anyon is cheaper than a pair of isolated e/3 anyons. Doping near the transition should in general lead to doping Cooper pairs and charge-2e superconductivity, with chiral edge modes of alternating central charge $c = \pm2$, which can coexist with translation symmetry breaking. Our framework unifies recent approaches to anyon superconductivity, reconciles it with strong repulsion and provides guidance for flat band moir\'e materials such as recent experiments in twisted MoTe$_2$. | [
{
"version": "v1",
"created": "Mon, 09 Jun 2025 17:59:13 GMT"
}
] | 2025-06-12T00:00:00 | [
[
"Pichler",
"Fabian",
""
],
[
"Kuhlenkamp",
"Clemens",
""
],
[
"Knap",
"Michael",
""
],
[
"Vishwanath",
"Ashvin",
""
]
] |
2506.08010 | Nick Jiang | Nick Jiang, Amil Dravid, Alexei Efros, Yossi Gandelsman | Vision Transformers Don't Need Trained Registers | Project page and code: https://avdravid.github.io/test-time-registers | null | null | null | cs.CV cs.AI | http://creativecommons.org/licenses/by/4.0/ | We investigate the mechanism underlying a previously identified phenomenon in Vision Transformers -- the emergence of high-norm tokens that lead to noisy attention maps. We observe that in multiple models (e.g., CLIP, DINOv2), a sparse set of neurons is responsible for concentrating high-norm activations on outlier tokens, leading to irregular attention patterns and degrading downstream visual processing. While the existing solution for removing these outliers involves retraining models from scratch with additional learned register tokens, we use our findings to create a training-free approach to mitigate these artifacts. By shifting the high-norm activations from our discovered register neurons into an additional untrained token, we can mimic the effect of register tokens on a model already trained without registers. We demonstrate that our method produces cleaner attention and feature maps, enhances performance over base models across multiple downstream visual tasks, and achieves results comparable to models explicitly trained with register tokens. We then extend test-time registers to off-the-shelf vision-language models to improve their interpretability. Our results suggest that test-time registers effectively take on the role of register tokens at test-time, offering a training-free solution for any pre-trained model released without them. | [
{
"version": "v1",
"created": "Mon, 09 Jun 2025 17:59:57 GMT"
},
{
"version": "v2",
"created": "Tue, 10 Jun 2025 22:38:32 GMT"
}
] | 2025-06-12T00:00:00 | [
[
"Jiang",
"Nick",
""
],
[
"Dravid",
"Amil",
""
],
[
"Efros",
"Alexei",
""
],
[
"Gandelsman",
"Yossi",
""
]
] |
2506.08022 | Chenxi Liu | Chenxi Liu, Tianyi Xiong, Ruibo Chen, Yihan Wu, Junfeng Guo, Tianyi Zhou, Heng Huang | Modality-Balancing Preference Optimization of Large Multimodal Models by Adversarial Negative Mining | null | null | null | null | cs.LG cs.AI cs.CL cs.CV | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | The task adaptation and alignment of Large Multimodal Models (LMMs) have been significantly advanced by instruction tuning and further strengthened by recent preference optimization. Yet, most LMMs still suffer from severe modality imbalance during reasoning, i.e., outweighing language prior biases over visual inputs, which bottlenecks their generalization to downstream tasks and causes hallucinations. However, existing preference optimization approaches for LMMs do not focus on restraining the internal biases of their Large Language Model (LLM) backbones when curating the training data. Moreover, they heavily rely on offline data and lack the capacity to explore diverse responses adaptive to dynamic distributional shifts during training. Meanwhile, Group Relative Policy Optimization (GRPO), a recent method using online-generated data and verified rewards to improve reasoning capabilities, remains largely underexplored in LMM alignment. In this paper, we propose a novel preference learning framework, Modality-Balancing Preference Optimization (MBPO), to address the modality imbalance in LMMs. MBPO constructs a more effective offline preference dataset by generating hard negatives, i.e., rejected responses misled by LLM biases due to limited usage of visual information, through adversarial perturbation of input images. Moreover, MBPO leverages the easy-to-verify nature of close-ended tasks to generate online responses with verified rewards. GRPO is then employed to train the model with offline-online hybrid data. Extensive experiments demonstrate that MBPO can enhance LMM performance on challenging vision-language tasks and effectively reduce hallucinations. | [
{
"version": "v1",
"created": "Tue, 20 May 2025 03:59:05 GMT"
},
{
"version": "v2",
"created": "Wed, 11 Jun 2025 03:54:30 GMT"
}
] | 2025-06-12T00:00:00 | [
[
"Liu",
"Chenxi",
""
],
[
"Xiong",
"Tianyi",
""
],
[
"Chen",
"Ruibo",
""
],
[
"Wu",
"Yihan",
""
],
[
"Guo",
"Junfeng",
""
],
[
"Zhou",
"Tianyi",
""
],
[
"Huang",
"Heng",
""
]
] |
2506.08033 | Axel Tahmasebimoradi | Axel TahmasebiMoradi, Vincent Ren, Benjamin Le-Creurer, Chetra Mang and Mouadh Yagoubi | Feasibility Study of CNNs and MLPs for Radiation Heat Transfer in 2-D Furnaces with Spectrally Participative Gases | null | null | null | null | eess.SY cs.LG cs.SY | http://creativecommons.org/licenses/by/4.0/ | Aiming to reduce the computational cost of numerical simulations, a convolutional neural network (CNN) and a multi-layer perceptron (MLP) are introduced to build a surrogate model to approximate radiative heat transfer solutions in a 2-D walled domain with participative gases. The originality of this work lays in the adaptation of the inputs of the problem (gas and wall properties) in order to fit with the CNN architecture, more commonly used for image processing. Two precision datasets have been created with the classical solver, ICARUS2D, that uses the discrete transfer radiation method with the statistical narrow bands model. The performance of the CNN architecture is compared to a more classical MLP architecture in terms of speed and accuracy. Thanks to Optuna, all results are obtained using the optimized hyper parameters networks. The results show a significant speedup with industrially acceptable relative errors compared to the classical solver for both architectures. Additionally, the CNN outperforms the MLP in terms of precision and is more robust and stable to changes in hyper-parameters. A performance analysis on the dataset size of the samples have also been carried out to gain a deeper understanding of the model behavior. | [
{
"version": "v1",
"created": "Mon, 02 Jun 2025 14:06:44 GMT"
},
{
"version": "v2",
"created": "Wed, 11 Jun 2025 10:14:50 GMT"
}
] | 2025-06-12T00:00:00 | [
[
"TahmasebiMoradi",
"Axel",
""
],
[
"Ren",
"Vincent",
""
],
[
"Le-Creurer",
"Benjamin",
""
],
[
"Mang",
"Chetra",
""
],
[
"Yagoubi",
"Mouadh",
""
]
] |
2506.08036 | Rahul Bhadani | Rahul Bhadani | Followstopper Revisited: Phase-space Lagrangian Controller for Traffic Decongestion | 10 figures, 11 pages | null | null | null | eess.SY cs.SY nlin.CD | http://creativecommons.org/licenses/by-nc-nd/4.0/ | This paper revisits Followerstopper, a phase-space-based control system that had demonstrated its ability to mitigate emergent traffic jams due to stop-and-go traffic during rush hour in the mixed-autonomy setting. Followerstopper was deployed on an autonomous vehicle. The controller attenuates the emanant traffic waves by regulating its velocity according to the relative distance and velocity of the leader car. While regulating the velocity, the controller also prevents the collision of the ego vehicle with the lead vehicle within the range specified by the controller's design parameter. The controller design is based on a configurable quadratic curve on relative distance-relative velocity phase-space that allows the transition of the regulated velocity from (i) no modification of input, (ii) decelerating to match the leader's velocity (iii) braking to avoid any imminent collision. In this paper, we explore the phase-space properties of Followerstopper and provide a detailed description of a nonlinear control law that regulates the reference input to Followerstopper within the physics-informed boundaries. We also provide a new discussion on the nominal control law that regulates the reference speed to Followerstopper to avoid unrealistic and unsafe acceleration. | [
{
"version": "v1",
"created": "Tue, 03 Jun 2025 23:07:42 GMT"
}
] | 2025-06-12T00:00:00 | [
[
"Bhadani",
"Rahul",
""
]
] |
2506.08048 | Zheng Han | Zheng Han, Jun Zhou, Jialun Pei, Jing Qin, Yingfang Fan, and Qi Dou | Toward Reliable AR-Guided Surgical Navigation: Interactive Deformation Modeling with Data-Driven Biomechanics and Prompts | null | null | 10.1109/TMI.2025.3577759 | null | cs.CV cs.AI cs.RO | http://creativecommons.org/licenses/by/4.0/ | In augmented reality (AR)-guided surgical navigation, preoperative organ models are superimposed onto the patient's intraoperative anatomy to visualize critical structures such as vessels and tumors. Accurate deformation modeling is essential to maintain the reliability of AR overlays by ensuring alignment between preoperative models and the dynamically changing anatomy. Although the finite element method (FEM) offers physically plausible modeling, its high computational cost limits intraoperative applicability. Moreover, existing algorithms often fail to handle large anatomical changes, such as those induced by pneumoperitoneum or ligament dissection, leading to inaccurate anatomical correspondences and compromised AR guidance. To address these challenges, we propose a data-driven biomechanics algorithm that preserves FEM-level accuracy while improving computational efficiency. In addition, we introduce a novel human-in-the-loop mechanism into the deformation modeling process. This enables surgeons to interactively provide prompts to correct anatomical misalignments, thereby incorporating clinical expertise and allowing the model to adapt dynamically to complex surgical scenarios. Experiments on a publicly available dataset demonstrate that our algorithm achieves a mean target registration error of 3.42 mm. Incorporating surgeon prompts through the interactive framework further reduces the error to 2.78 mm, surpassing state-of-the-art methods in volumetric accuracy. These results highlight the ability of our framework to deliver efficient and accurate deformation modeling while enhancing surgeon-algorithm collaboration, paving the way for safer and more reliable computer-assisted surgeries. | [
{
"version": "v1",
"created": "Sun, 08 Jun 2025 14:19:54 GMT"
},
{
"version": "v2",
"created": "Wed, 11 Jun 2025 03:34:08 GMT"
}
] | 2025-06-12T00:00:00 | [
[
"Han",
"Zheng",
""
],
[
"Zhou",
"Jun",
""
],
[
"Pei",
"Jialun",
""
],
[
"Qin",
"Jing",
""
],
[
"Fan",
"Yingfang",
""
],
[
"Dou",
"Qi",
""
]
] |
2506.08049 | Tengfei Lyu | Tengfei Lyu, Weijia Zhang, Hao Liu | Physics-Informed Teleconnection-Aware Transformer for Global Subseasonal-to-Seasonal Forecasting | null | null | null | null | stat.ML cs.AI cs.LG | http://creativecommons.org/licenses/by/4.0/ | Subseasonal-to-seasonal (S2S) forecasting, which predicts climate conditions from several weeks to months in advance, presents significant challenges due to the chaotic dynamics of atmospheric systems and complex interactions across multiple scales. Current approaches often fail to explicitly model underlying physical processes and teleconnections that are crucial at S2S timescales. We introduce TelePiT, a novel deep learning architecture that enhances global S2S forecasting through integrated multi-scale physics and teleconnection awareness. Our approach consists of three key components: (1) Spherical Harmonic Embedding, which accurately encodes global atmospheric variables onto spherical geometry; (2) Multi-Scale Physics-Informed Neural ODE, which explicitly captures atmospheric physical processes across multiple learnable frequency bands; (3) Teleconnection-Aware Transformer, which models critical global climate interactions through tactfully injecting teleconnection patterns into the self-attention. Extensive experiments demonstrate that TelePiT significantly outperforms state-of-the-art data-driven baselines and operational numerical weather prediction systems, with remarkable improvements for atmospheric variables including a 57.7% reduction in RMSE for 2-meter temperature compared to previous best models. | [
{
"version": "v1",
"created": "Sun, 08 Jun 2025 16:32:21 GMT"
},
{
"version": "v2",
"created": "Wed, 11 Jun 2025 10:52:56 GMT"
}
] | 2025-06-12T00:00:00 | [
[
"Lyu",
"Tengfei",
""
],
[
"Zhang",
"Weijia",
""
],
[
"Liu",
"Hao",
""
]
] |
2506.08054 | Yiming Wang | Yiming Wang, Hao Peng, Senzhang Wang, Haohua Du, Chunyang Liu, Jia Wu, Guanlin Wu | STAMImputer: Spatio-Temporal Attention MoE for Traffic Data Imputation | 10 pages, 5 figures, 3 tables. Extended version of paper accepted at IJCAI 2025 | null | null | null | cs.LG cs.AI | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Traffic data imputation is fundamentally important to support various applications in intelligent transportation systems such as traffic flow prediction. However, existing time-to-space sequential methods often fail to effectively extract features in block-wise missing data scenarios. Meanwhile, the static graph structure for spatial feature propagation significantly constrains the models flexibility in handling the distribution shift issue for the nonstationary traffic data. To address these issues, this paper proposes a SpatioTemporal Attention Mixture of experts network named STAMImputer for traffic data imputation. Specifically, we introduce a Mixture of Experts (MoE) framework to capture latent spatio-temporal features and their influence weights, effectively imputing block missing. A novel Low-rank guided Sampling Graph ATtention (LrSGAT) mechanism is designed to dynamically balance the local and global correlations across road networks. The sampled attention vectors are utilized to generate dynamic graphs that capture real-time spatial correlations. Extensive experiments are conducted on four traffic datasets for evaluation. The result shows STAMImputer achieves significantly performance improvement compared with existing SOTA approaches. Our codes are available at https://github.com/RingBDStack/STAMImupter. | [
{
"version": "v1",
"created": "Mon, 09 Jun 2025 04:05:00 GMT"
},
{
"version": "v2",
"created": "Wed, 11 Jun 2025 02:33:59 GMT"
}
] | 2025-06-12T00:00:00 | [
[
"Wang",
"Yiming",
""
],
[
"Peng",
"Hao",
""
],
[
"Wang",
"Senzhang",
""
],
[
"Du",
"Haohua",
""
],
[
"Liu",
"Chunyang",
""
],
[
"Wu",
"Jia",
""
],
[
"Wu",
"Guanlin",
""
]
] |
2506.08065 | Ye Zhu | Ye Zhu, Duo Xu, Zhiwei Deng, Jonathan C. Tan, Olga Russakovsky | Dynamic Diffusion Schr\"odinger Bridge in Astrophysical Observational Inversions | Preprint. Code will be available at https://github.com/L-YeZhu/AstroDSB | null | null | null | astro-ph.IM cs.LG | http://creativecommons.org/licenses/by/4.0/ | We study Diffusion Schr\"odinger Bridge (DSB) models in the context of dynamical astrophysical systems, specifically tackling observational inverse prediction tasks within Giant Molecular Clouds (GMCs) for star formation. We introduce the Astro-DSB model, a variant of DSB with the pairwise domain assumption tailored for astrophysical dynamics. By investigating its learning process and prediction performance in both physically simulated data and in real observations (the Taurus B213 data), we present two main takeaways. First, from the astrophysical perspective, our proposed paired DSB method improves interpretability, learning efficiency, and prediction performance over conventional astrostatistical and other machine learning methods. Second, from the generative modeling perspective, probabilistic generative modeling reveals improvements over discriminative pixel-to-pixel modeling in Out-Of-Distribution (OOD) testing cases of physical simulations with unseen initial conditions and different dominant physical processes. Our study expands research into diffusion models beyond the traditional visual synthesis application and provides evidence of the models' learning abilities beyond pure data statistics, paving a path for future physics-aware generative models which can align dynamics between machine learning and real (astro)physical systems. | [
{
"version": "v1",
"created": "Mon, 09 Jun 2025 12:35:36 GMT"
},
{
"version": "v2",
"created": "Wed, 11 Jun 2025 16:43:38 GMT"
}
] | 2025-06-12T00:00:00 | [
[
"Zhu",
"Ye",
""
],
[
"Xu",
"Duo",
""
],
[
"Deng",
"Zhiwei",
""
],
[
"Tan",
"Jonathan C.",
""
],
[
"Russakovsky",
"Olga",
""
]
] |
2506.08068 | Paolo Cifani | Paolo Cifani, Franco Flandoli | Diffusion properties of small-scale fractional transport models | null | null | null | null | cond-mat.stat-mech | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Stochastic transport due to a velocity field modeled by the superposition of small-scale divergence free vector fields activated by Fractional Gaussian Noises (FGN) is numerically investigated. We present two non-trivial contributions: the first one is the definition of a model where different space-time structures can be compared on the same ground: this is achieved by imposing the same average kinetic energy to a standard Ornstein-Uhlenbeck approximation, then taking the limit to the idealized white noise structure. The second contribution, based on the previous one, is the discover that a mixing spatial structure with persistent FGN in the Fourier components induces a classical Brownian diffusion of passive particles, with suitable diffusion coefficient; namely, the memory of FGN is lost in the space complexity of the velocity field. | [
{
"version": "v1",
"created": "Mon, 09 Jun 2025 15:15:43 GMT"
},
{
"version": "v2",
"created": "Wed, 11 Jun 2025 07:40:24 GMT"
}
] | 2025-06-12T00:00:00 | [
[
"Cifani",
"Paolo",
""
],
[
"Flandoli",
"Franco",
""
]
] |
2506.08079 | Daniel Horta Darrington Dr | Danny Horta, Melissa K. Ness | Hydrostatic and explosive $\alpha$-element chemical abundances of Milky Way globular clusters, halo substructures, and satellite galaxies | Under review in MNRAS. 6 pages, 5 figures. Comments welcome! | null | null | null | astro-ph.GA | http://creativecommons.org/licenses/by/4.0/ | Stellar atmospheric element abundance ratios of stars retain information about their birth conditions, helping elucidate their origin and nature. In this letter, we analyse and contrast the hydrostatic and explosive $\alpha$-element abundance ratios, and the ratio of the two (the hex ratio), for a large sample of Galactic globular clusters (GCs), halo substructures, satellite galaxies, and the Milky Way high-/low-$\alpha$ discs using data from the $APOGEE$ survey. Our results show that: $i$) Milky Way GCs and halo substructures appear to have qualitatively similar hex ratios across a broad range of [Fe/H], that are higher than that of dwarf satellite galaxies of similar [Fe/H]; $ii$) for all stellar populations studied, there is a trend in the hex ratio with [Fe/H]; $iii$) there is a weak trend in the hex ratio with respect to age for Galactic GCs, but not with initial or final GC mass; $iv$) there are no differences in the hex ratio between GCs formed $in$ $situ$ versus those labelled as accreted. | [
{
"version": "v1",
"created": "Mon, 09 Jun 2025 18:00:01 GMT"
},
{
"version": "v2",
"created": "Wed, 11 Jun 2025 07:10:13 GMT"
}
] | 2025-06-12T00:00:00 | [
[
"Horta",
"Danny",
""
],
[
"Ness",
"Melissa K.",
""
]
] |
2506.08109 | Aritra Sanyal | Aritra Sanyal | Cosmic Hysteresis in $f(R)$ Gravity: A Thermodynamic Perspective on Cyclic Bouncing Universes | The manuscript is being withdrawn due to the need for substantial revisions based on new insights and improved analysis. A significantly revised version will be submitted in the future | null | null | null | gr-qc | http://creativecommons.org/licenses/by/4.0/ | We investigate the emergence of cosmic hysteresis in cyclic bouncing cosmologies within the framework of $f(R)$ modified gravity theories. Building upon previous studies that explored hysteresis phenomena in braneworld and Einstein-Gauss-Bonnet gravity, we analyze how scalar field dynamics coupled to a modified gravitational background generates asymmetric pressure evolution during expansion and contraction phases. This asymmetry manifests as a non-vanishing work integral over complete cosmic cycles, $\oint p\,dV \neq 0$, which we interpret as a fundamental thermodynamic signature of irreversible cyclic evolution. We derive the modified Friedmann equations for a representative $f(R) = R + \alpha R^2$ model, establish analytical conditions for bounce and turnaround events, and quantify the thermodynamic work performed during each cycle through comprehensive numerical analysis. Our findings demonstrate that hysteresis effects are generic features of scalar-tensor dynamics in higher-curvature gravity theories and can drive cumulative evolution of the cosmic scale factor across multiple cycles, potentially providing new insights into the resolution of cosmological singularities and the thermodynamic arrow of time in modified gravity frameworks. | [
{
"version": "v1",
"created": "Mon, 09 Jun 2025 18:05:48 GMT"
},
{
"version": "v2",
"created": "Wed, 11 Jun 2025 16:09:39 GMT"
}
] | 2025-06-12T00:00:00 | [
[
"Sanyal",
"Aritra",
""
]
] |
2506.08128 | Sudip Chakrabarty | Sudip Chakrabarty, Amit Kundu, A. S. Majumdar | Network nonlocality with continuous variables: probing nonbilocality through pseudospin measurements | 15 pages, 9 figures | null | null | null | quant-ph | http://creativecommons.org/licenses/by/4.0/ | Quantum networks offer a compelling platform for probing nonlocal correlations beyond the Bell scenario. While network nonlocality has been well studied in discrete-variable (DV) systems, its exploration in continuous-variable (CV) systems remains limited. In this work, we study the bilocal scenario, the simplest quantum network, in a CV framework using pseudospin measurements. We show that two-mode squeezed vacuum (TMSV) states violate the bilocality inequality for any nonzero squeezing, with the violation increasing monotonically and reaching the maximum in the infinite squeezing limit. The robustness of this violation is explored in the presence of local thermal noise. Next, we demonstrate nonbilocality in two key non-Gaussian states, viz., the entangled coherent state and the CV Werner state. Finally, we investigate the enhancement of nonbilocal correlations through photon subtraction applied to the TMSV state. We show that a coherent superposition of single-photon subtractions across modes enables maximal bilocality violation even at zero squeezing, demonstrating non-Gaussianity as a strong resource in the low-squeezing regime. Our findings underscore the potential of CV systems for realizing network-based quantum correlations. | [
{
"version": "v1",
"created": "Mon, 09 Jun 2025 18:31:17 GMT"
},
{
"version": "v2",
"created": "Wed, 11 Jun 2025 10:33:33 GMT"
}
] | 2025-06-12T00:00:00 | [
[
"Chakrabarty",
"Sudip",
""
],
[
"Kundu",
"Amit",
""
],
[
"Majumdar",
"A. S.",
""
]
] |
2506.08137 | Oishee Bintey Hoque | Oishee Bintey Hoque, Abhijin Adiga, Aniruddha Adiga, Siddharth Chaudhary, Madhav V. Marathe, S. S. Ravi, Kirti Rajagopalan, Amanda Wilson and Samarth Swarup | IGraSS: Learning to Identify Infrastructure Networks from Satellite Imagery by Iterative Graph-constrained Semantic Segmentation | null | null | null | null | cs.CV cs.AI | http://creativecommons.org/licenses/by/4.0/ | Accurate canal network mapping is essential for water management, including irrigation planning and infrastructure maintenance. State-of-the-art semantic segmentation models for infrastructure mapping, such as roads, rely on large, well-annotated remote sensing datasets. However, incomplete or inadequate ground truth can hinder these learning approaches. Many infrastructure networks have graph-level properties such as reachability to a source (like canals) or connectivity (roads) that can be leveraged to improve these existing ground truth. This paper develops a novel iterative framework IGraSS, combining a semantic segmentation module-incorporating RGB and additional modalities (NDWI, DEM)-with a graph-based ground-truth refinement module. The segmentation module processes satellite imagery patches, while the refinement module operates on the entire data viewing the infrastructure network as a graph. Experiments show that IGraSS reduces unreachable canal segments from around 18% to 3%, and training with refined ground truth significantly improves canal identification. IGraSS serves as a robust framework for both refining noisy ground truth and mapping canal networks from remote sensing imagery. We also demonstrate the effectiveness and generalizability of IGraSS using road networks as an example, applying a different graph-theoretic constraint to complete road networks. | [
{
"version": "v1",
"created": "Mon, 09 Jun 2025 18:40:22 GMT"
},
{
"version": "v2",
"created": "Wed, 11 Jun 2025 02:45:23 GMT"
}
] | 2025-06-12T00:00:00 | [
[
"Hoque",
"Oishee Bintey",
""
],
[
"Adiga",
"Abhijin",
""
],
[
"Adiga",
"Aniruddha",
""
],
[
"Chaudhary",
"Siddharth",
""
],
[
"Marathe",
"Madhav V.",
""
],
[
"Ravi",
"S. S.",
""
],
[
"Rajagopalan",
"Kirti",
""
],
[
"Wilson",
"Amanda",
""
],
[
"Swarup",
"Samarth",
""
]
] |
2506.08174 | Pedro Brom | Li Weigang and Pedro Carvalho Brom | LLM-BT-Terms: Back-Translation as a Framework for Terminology Standardization and Dynamic Semantic Embedding | 23 pages | null | null | null | cs.CL | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | The rapid expansion of English technical terminology presents a significant challenge to traditional expert-based standardization, particularly in rapidly developing areas such as artificial intelligence and quantum computing. Manual approaches face difficulties in maintaining consistent multilingual terminology. To address this, we introduce LLM-BT, a back-translation framework powered by large language models (LLMs) designed to automate terminology verification and standardization through cross-lingual semantic alignment. Our key contributions include: (1) term-level consistency validation: by performing English -> intermediate language -> English back-translation, LLM-BT achieves high term consistency across different models (such as GPT-4, DeepSeek, and Grok). Case studies demonstrate over 90 percent of terms are preserved either exactly or semantically; (2) multi-path verification workflow: we develop a novel pipeline described as Retrieve -> Generate -> Verify -> Optimize, which supports both serial paths (e.g., English -> Simplified Chinese -> Traditional Chinese -> English) and parallel paths (e.g., English -> Chinese / Portuguese -> English). BLEU scores and term-level accuracy indicate strong cross-lingual robustness, with BLEU scores exceeding 0.45 and Portuguese term accuracy reaching 100 percent; (3) back-translation as semantic embedding: we reinterpret back-translation as a form of dynamic semantic embedding that uncovers latent trajectories of meaning. In contrast to static embeddings, LLM-BT offers transparent, path-based embeddings shaped by the evolution of the models. This reframing positions back-translation as an active mechanism for multilingual terminology standardization, fostering collaboration between machines and humans - machines preserve semantic integrity, while humans provide cultural interpretation. | [
{
"version": "v1",
"created": "Mon, 09 Jun 2025 19:39:09 GMT"
},
{
"version": "v2",
"created": "Wed, 11 Jun 2025 17:04:39 GMT"
}
] | 2025-06-12T00:00:00 | [
[
"Weigang",
"Li",
""
],
[
"Brom",
"Pedro Carvalho",
""
]
] |
2506.08184 | Chupei Wang | Chupei Wang (University of Virginia), Jiaqiu Vince Sun (New York University) | Unable to Forget: Proactive lnterference Reveals Working Memory Limits in LLMs Beyond Context Length | null | null | null | null | cs.CL cs.AI q-bio.NC | http://creativecommons.org/licenses/by/4.0/ | Information retrieval in Large Language Models (LLMs) is increasingly recognized as intertwined with generation capabilities rather than mere lookup. While longer contexts are often assumed to improve retrieval, the effects of intra-context interference remain understudied. To address this, we adapt the proactive interference (PI) paradigm from cognitive science, where earlier information disrupts recall of newer updates. In humans, susceptibility to such interference is inversely linked to working memory capacity. We introduce PI-LLM, an evaluation that sequentially streams semantically related key-value updates and queries only the final values. Although these final values are clearly positioned just before the query, LLM retrieval accuracy declines log-linearly toward zero as interference accumulates; errors arise from retrieving previously overwritten values. Attempts to mitigate interference via prompt engineering (e.g., instructing models to ignore earlier input) yield limited success. These findings reveal a fundamental constraint on LLMs' ability to disentangle interference and flexibly manipulate information, suggesting a working memory bottleneck beyond mere context access. This calls for approaches that strengthen models' ability to suppress irrelevant content during retrieval. | [
{
"version": "v1",
"created": "Mon, 09 Jun 2025 19:49:11 GMT"
},
{
"version": "v2",
"created": "Wed, 11 Jun 2025 09:53:49 GMT"
}
] | 2025-06-12T00:00:00 | [
[
"Wang",
"Chupei",
"",
"University of Virginia"
],
[
"Sun",
"Jiaqiu Vince",
"",
"New York University"
]
] |
2506.08194 | Mateusz Michalkiewicz | Mateusz Michalkiewicz and Anekha Sokhal and Tadeusz Michalkiewicz and Piotr Pawlikowski and Mahsa Baktashmotlagh and Varun Jampani and Guha Balakrishnan | GIQ: Benchmarking 3D Geometric Reasoning of Vision Foundation Models with Simulated and Real Polyhedra | 15 pages, 4 figures | null | null | null | cs.CV | http://creativecommons.org/licenses/by/4.0/ | Monocular 3D reconstruction methods and vision-language models (VLMs) demonstrate impressive results on standard benchmarks, yet their true understanding of geometric properties remains unclear. We introduce GIQ , a comprehensive benchmark specifically designed to evaluate the geometric reasoning capabilities of vision and vision-language foundation models. GIQ comprises synthetic and real-world images of 224 diverse polyhedra - including Platonic, Archimedean, Johnson, and Catalan solids, as well as stellations and compound shapes - covering varying levels of complexity and symmetry. Through systematic experiments involving monocular 3D reconstruction, 3D symmetry detection, mental rotation tests, and zero-shot shape classification tasks, we reveal significant shortcomings in current models. State-of-the-art reconstruction algorithms trained on extensive 3D datasets struggle to reconstruct even basic geometric forms accurately. While foundation models effectively detect specific 3D symmetry elements via linear probing, they falter significantly in tasks requiring detailed geometric differentiation, such as mental rotation. Moreover, advanced vision-language assistants exhibit remarkably low accuracy on complex polyhedra, systematically misinterpreting basic properties like face geometry, convexity, and compound structures. GIQ is publicly available, providing a structured platform to highlight and address critical gaps in geometric intelligence, facilitating future progress in robust, geometry-aware representation learning. | [
{
"version": "v1",
"created": "Mon, 09 Jun 2025 20:11:21 GMT"
},
{
"version": "v2",
"created": "Wed, 11 Jun 2025 02:23:29 GMT"
}
] | 2025-06-12T00:00:00 | [
[
"Michalkiewicz",
"Mateusz",
""
],
[
"Sokhal",
"Anekha",
""
],
[
"Michalkiewicz",
"Tadeusz",
""
],
[
"Pawlikowski",
"Piotr",
""
],
[
"Baktashmotlagh",
"Mahsa",
""
],
[
"Jampani",
"Varun",
""
],
[
"Balakrishnan",
"Guha",
""
]
] |
2506.08213 | Jasem Hamoud Mohamad | Jasem Hamoud, Artem Kornosov | Degree Sequence of Albertson and $\sigma$-Indices on Trees of Order $n\geqslant 3$ | 11 pages, 4 figures, 1 table, $67^{th}$ All-Russian Scientific Conference of the Moscow Institute of Physics and Technology, Comments welcome! arXiv admin note: text overlap with arXiv:2503.12909 | null | null | null | math.CO | http://creativecommons.org/licenses/by/4.0/ | In this paper, we presented a study of topological indices on trees, where we show a relationship with irregularity of Albertson index and minimum, maximum degrees $\delta,\Delta$ of graph $G$, where contribute vital roles in determining connection, shading, component incorporation, and realisability where well-known Albertson index as: $\operatorname{irr}(G)=\sum_{uv\in E(G)}\lvert d_u(G)-d_v(G) \rvert$. The sigma index on trees that we introduced as $\sigma(T)=(d_1-1)^3+\sum_{i=1}^{3}(d_i-1)(d_i-2)+(d_3-1)^3$. | [
{
"version": "v1",
"created": "Mon, 09 Jun 2025 20:31:29 GMT"
},
{
"version": "v2",
"created": "Wed, 11 Jun 2025 07:12:30 GMT"
}
] | 2025-06-12T00:00:00 | [
[
"Hamoud",
"Jasem",
""
],
[
"Kornosov",
"Artem",
""
]
] |
2506.08242 | Fabio Frommer | Fabio Frommer | The Kirkwood closure point process: A solution of the Kirkwood-Salsburg equations for negative activities | null | null | null | null | math-ph math.MP math.PR | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | The Kirkwood superposition is a well-known tool in statistical physics to approximate the $n$-point correlation functions for $n\geq 3$ in terms of the density $\rho $ and the radial distribution function $g$ of the underlying system. However, it is unclear whether these approximations are themselves the correlation functions of some point process. If they are, this process is called the Kirkwood closure process. For the case that $g$ is the negative exponential of some nonnegative and regular pair potential $u$ existence of the the Kirkwood closure process was proved by Ambartzumian and Sukiasian. This result was generalized to the case that $u$ is a locally stable and regular pair potential by Kuna, Lebowitz and Speer, provided that $\rho$ is sufficiently small. In this work, it is shown that it suffices for $u$ to be stable and regular to ensure the existence of the Kirkwood closure process. Furthermore, for locally stable $u$ it is proved that the Kirkwood closure process is Gibbs and that the kernel of the GNZ-equation satisfies a Kirkwood-Salsburg type equation. | [
{
"version": "v1",
"created": "Mon, 09 Jun 2025 21:20:42 GMT"
},
{
"version": "v2",
"created": "Wed, 11 Jun 2025 11:20:09 GMT"
}
] | 2025-06-12T00:00:00 | [
[
"Frommer",
"Fabio",
""
]
] |
2506.08247 | Saulius Vaitiek\.enas | J. Zhao, A. Mazanik, D. Razmadze, Y. Liu, P. Krogstrup, F. S. Bergeret, and S. Vaitiek\.enas | Spin-split superconductivity in spin-orbit coupled hybrid nanowires with ferromagnetic barriers | 10 pages, 5+5 figures | null | null | NBI QDEV 2025 | cond-mat.mes-hall cond-mat.supr-con | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | We report transport studies of hybrid Josephson junctions based on semiconducting InAs nanowires with fully overlapping epitaxial ferromagnetic insulator EuS and superconducting Al partial shells. Current-biased measurements reveal a hysteretic superconducting window with a sizable supercurrent near the coercive field of the ferromagnetic insulator, accompanied by multiple Andreev reflections. Tunneling spectroscopy shows a superconducting gap characterized by three peaks, which we attribute to tunneling between exchange-split superconductors. A theoretical model reproduces the observed features and indicates that spin mixing, driven by sizable spin-orbit coupling, is essential to their formation. Our results demonstrate proximity-induced superconductivity through a ferromagnetic insulator and establish a new platform for exploring spin-triplet pairing. | [
{
"version": "v1",
"created": "Mon, 09 Jun 2025 21:29:24 GMT"
},
{
"version": "v2",
"created": "Wed, 11 Jun 2025 17:05:14 GMT"
}
] | 2025-06-12T00:00:00 | [
[
"Zhao",
"J.",
""
],
[
"Mazanik",
"A.",
""
],
[
"Razmadze",
"D.",
""
],
[
"Liu",
"Y.",
""
],
[
"Krogstrup",
"P.",
""
],
[
"Bergeret",
"F. S.",
""
],
[
"Vaitiekėnas",
"S.",
""
]
] |
2506.08248 | Sebasti\'an Saldivia | Sebasti\'an Saldivia, Felipe A. Asenjo and Pablo S. Moya | The effect of plasma expansion on the dispersion properties of MHD waves | 19 pages, 4 figures. Submitted to Physica Scripta, Focus Edition: Plasma Physics in Latin America | null | null | null | physics.plasm-ph astro-ph.SR | http://creativecommons.org/licenses/by-nc-nd/4.0/ | In this work, we employ the set of ideal expanding magnetohydrodynamic (MHD) equations within the Expanding Box Model (EBM) framework to theoretically characterize the effects of radial solar wind expansion on its characteristic linear MHD waves. Through the analytical derivation of dispersion relations by a first-order expansion of the MHD-EBM equations, we explore the changes in wave propagation across a range of heliocentric distances on the linear magnetohydrodynamic modes: the Alfv\'en mode and the fast and slow magnetosonic modes, as obtained from the ideal MHD-EBM equations. Our findings reveal a spatial dependence in the derived dispersion relations that aligns with both the literature and the traditional ideal MHD case in the non-expanding limit, thereby helping to bridge the gap between theory and observation in solar wind dynamics. We observe a general decrease in wave frequencies as the plasma expands farther from the Sun. This decrease is reflected in the dispersion relations through the radial decrease of both the Alfv\'en and sound speeds, which decrease proportionally to $1/R$ and $1/R^{\gamma - 1}$, respectively, where $\gamma$ is the plasma polytropic index. The fast magnetosonic mode frequency and phase speed are significantly affected by the polytropic index value. We consider three models for the polytropic index evolution in the expanding solar wind: a constant (quasi-adiabatic) case, a radially decreasing profile in the outer heliosphere, and a model incorporating thermodynamic heating effects. Notably, we find that in the case of a decreasing polytropic index, the fast magnetosonic mode experiences an acceleration in the distant heliosphere, highlighting the significant influence of expansion on solar wind dynamics. | [
{
"version": "v1",
"created": "Mon, 09 Jun 2025 21:32:25 GMT"
},
{
"version": "v2",
"created": "Wed, 11 Jun 2025 03:36:26 GMT"
}
] | 2025-06-12T00:00:00 | [
[
"Saldivia",
"Sebastián",
""
],
[
"Asenjo",
"Felipe A.",
""
],
[
"Moya",
"Pablo S.",
""
]
] |
2506.08264 | Jan Verlage | Jan Verlage and Peter Kratzer | Bound States at Semiconductor -- Mott Insulator Interfaces | null | null | null | null | cond-mat.str-el | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Utilizing the hierarchy of correlations in the context of a Fermi-Hubbard model, we deduce the presence of quasi-particle bound states at the interface between a Mott insulator and a semiconductor, as well as within a semiconductor-Mott-semiconductor heterostructure forming a quantum well. In the case of the solitary interface, the existence of bound states necessitates the presence of an additional perturbation with a minimal strength depending on the spin background of the Mott insulator. Conversely, within the quantum well, this additional perturbation is still required to have bound states while standing-wave solutions even exist in its absence. | [
{
"version": "v1",
"created": "Mon, 09 Jun 2025 21:59:22 GMT"
},
{
"version": "v2",
"created": "Wed, 11 Jun 2025 09:14:17 GMT"
}
] | 2025-06-12T00:00:00 | [
[
"Verlage",
"Jan",
""
],
[
"Kratzer",
"Peter",
""
]
] |
2506.08270 | Mansooreh Montazerin | Zitong Huang, Mansooreh Montazerin and Ajitesh Srivastava | SWAT-NN: Simultaneous Weights and Architecture Training for Neural Networks in a Latent Space | null | null | null | null | cs.LG | http://creativecommons.org/licenses/by/4.0/ | Designing neural networks typically relies on manual trial and error or a neural architecture search (NAS) followed by weight training. The former is time-consuming and labor-intensive, while the latter often discretizes architecture search and weight optimization. In this paper, we propose a fundamentally different approach that simultaneously optimizes both the architecture and the weights of a neural network. Our framework first trains a universal multi-scale autoencoder that embeds both architectural and parametric information into a continuous latent space, where functionally similar neural networks are mapped closer together. Given a dataset, we then randomly initialize a point in the embedding space and update it via gradient descent to obtain the optimal neural network, jointly optimizing its structure and weights. The optimization process incorporates sparsity and compactness penalties to promote efficient models. Experiments on synthetic regression tasks demonstrate that our method effectively discovers sparse and compact neural networks with strong performance. | [
{
"version": "v1",
"created": "Mon, 09 Jun 2025 22:22:37 GMT"
},
{
"version": "v2",
"created": "Wed, 11 Jun 2025 01:56:59 GMT"
}
] | 2025-06-12T00:00:00 | [
[
"Huang",
"Zitong",
""
],
[
"Montazerin",
"Mansooreh",
""
],
[
"Srivastava",
"Ajitesh",
""
]
] |
2506.08296 | Heng Zhang | Hongjun Wu, Heng Zhang, Pengsong Zhang, Jin Wang, Cong Wang | HiBerNAC: Hierarchical Brain-emulated Robotic Neural Agent Collective for Disentangling Complex Manipulation | 31 pages,5 figures | null | null | null | cs.RO | http://creativecommons.org/licenses/by/4.0/ | Recent advances in multimodal vision-language-action (VLA) models have revolutionized traditional robot learning, enabling systems to interpret vision, language, and action in unified frameworks for complex task planning. However, mastering complex manipulation tasks remains an open challenge, constrained by limitations in persistent contextual memory, multi-agent coordination under uncertainty, and dynamic long-horizon planning across variable sequences. To address this challenge, we propose \textbf{HiBerNAC}, a \textbf{Hi}erarchical \textbf{B}rain-\textbf{e}mulated \textbf{r}obotic \textbf{N}eural \textbf{A}gent \textbf{C}ollective, inspired by breakthroughs in neuroscience, particularly in neural circuit mechanisms and hierarchical decision-making. Our framework combines: (1) multimodal VLA planning and reasoning with (2) neuro-inspired reflection and multi-agent mechanisms, specifically designed for complex robotic manipulation tasks. By leveraging neuro-inspired functional modules with decentralized multi-agent collaboration, our approach enables robust and enhanced real-time execution of complex manipulation tasks. In addition, the agentic system exhibits scalable collective intelligence via dynamic agent specialization, adapting its coordination strategy to variable task horizons and complexity. Through extensive experiments on complex manipulation tasks compared with state-of-the-art VLA models, we demonstrate that \textbf{HiBerNAC} reduces average long-horizon task completion time by 23\%, and achieves non-zero success rates (12\textendash 31\%) on multi-path tasks where prior state-of-the-art VLA models consistently fail. These results provide indicative evidence for bridging biological cognition and robotic learning mechanisms. | [
{
"version": "v1",
"created": "Mon, 09 Jun 2025 23:59:31 GMT"
},
{
"version": "v2",
"created": "Wed, 11 Jun 2025 14:29:23 GMT"
}
] | 2025-06-12T00:00:00 | [
[
"Wu",
"Hongjun",
""
],
[
"Zhang",
"Heng",
""
],
[
"Zhang",
"Pengsong",
""
],
[
"Wang",
"Jin",
""
],
[
"Wang",
"Cong",
""
]
] |
2506.08309 | Katherine Tieu | Katherine Tieu, Dongqi Fu, Zihao Li, Ross Maciejewski, Jingrui He | Learnable Spatial-Temporal Positional Encoding for Link Prediction | Accepted by ICML 2025. 28 pages, 1 figures, 22 tables | null | null | null | cs.LG cs.AI | http://creativecommons.org/licenses/by-nc-sa/4.0/ | Accurate predictions rely on the expressiveness power of graph deep learning frameworks like graph neural networks and graph transformers, where a positional encoding mechanism has become much more indispensable in recent state-of-the-art works to record the canonical position information. However, the current positional encoding is limited in three aspects: (1) most positional encoding methods use pre-defined, and fixed functions, which are inadequate to adapt to the complex attributed graphs; (2) a few pioneering works proposed the learnable positional encoding but are still limited to the structural information, not considering the real-world time-evolving topological and feature information; (3) most positional encoding methods are equipped with transformers' attention mechanism to fully leverage their capabilities, where the dense or relational attention is often unaffordable on large-scale structured data. Hence, we aim to develop Learnable Spatial-Temporal Positional Encoding in an effective and efficient manner and propose a simple temporal link prediction model named L-STEP. Briefly, for L-STEP, we (1) prove the proposed positional learning scheme can preserve the graph property from the spatial-temporal spectral viewpoint, (2) verify that MLPs can fully exploit the expressiveness and reach transformers' performance on that encoding, (3) change different initial positional encoding inputs to show robustness, (4) analyze the theoretical complexity and obtain less empirical running time than SOTA, and (5) demonstrate its temporal link prediction out-performance on 13 classic datasets and with 10 algorithms in both transductive and inductive settings using 3 different sampling strategies. Also, L-STEP obtains the leading performance in the newest large-scale TGB benchmark. Our code is available at https://github.com/kthrn22/L-STEP. | [
{
"version": "v1",
"created": "Tue, 10 Jun 2025 00:35:53 GMT"
},
{
"version": "v2",
"created": "Wed, 11 Jun 2025 03:31:52 GMT"
}
] | 2025-06-12T00:00:00 | [
[
"Tieu",
"Katherine",
""
],
[
"Fu",
"Dongqi",
""
],
[
"Li",
"Zihao",
""
],
[
"Maciejewski",
"Ross",
""
],
[
"He",
"Jingrui",
""
]
] |
2506.08324 | Guandong Li | Guandong Li, Mengxia Ye | Hyperspectral Image Classification via Transformer-based Spectral-Spatial Attention Decoupling and Adaptive Gating | arXiv admin note: substantial text overlap with arXiv:2504.15155, arXiv:2504.13045, arXiv:2503.23472 | null | null | null | cs.CV | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Deep neural networks face several challenges in hyperspectral image classification, including high-dimensional data, sparse distribution of ground objects, and spectral redundancy, which often lead to classification overfitting and limited generalization capability. To more effectively extract and fuse spatial context with fine spectral information in hyperspectral image (HSI) classification, this paper proposes a novel network architecture called STNet. The core advantage of STNet stems from the dual innovative design of its Spatial-Spectral Transformer module: first, the fundamental explicit decoupling of spatial and spectral attention ensures targeted capture of key information in HSI; second, two functionally distinct gating mechanisms perform intelligent regulation at both the fusion level of attention flows (adaptive attention fusion gating) and the internal level of feature transformation (GFFN). This characteristic demonstrates superior feature extraction and fusion capabilities compared to traditional convolutional neural networks, while reducing overfitting risks in small-sample and high-noise scenarios. STNet enhances model representation capability without increasing network depth or width. The proposed method demonstrates superior performance on IN, UP, and KSC datasets, outperforming mainstream hyperspectral image classification approaches. | [
{
"version": "v1",
"created": "Tue, 10 Jun 2025 01:24:35 GMT"
},
{
"version": "v2",
"created": "Wed, 11 Jun 2025 01:30:34 GMT"
}
] | 2025-06-12T00:00:00 | [
[
"Li",
"Guandong",
""
],
[
"Ye",
"Mengxia",
""
]
] |
2506.08336 | Changjiang Li | Li Changjiang, Liang Jiacheng, Cao Bochuan, Chen Jinghui, Wang Ting | Your Agent Can Defend Itself against Backdoor Attacks | null | null | null | null | cs.CR cs.AI cs.LG | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Despite their growing adoption across domains, large language model (LLM)-powered agents face significant security risks from backdoor attacks during training and fine-tuning. These compromised agents can subsequently be manipulated to execute malicious operations when presented with specific triggers in their inputs or environments. To address this pressing risk, we present ReAgent, a novel defense against a range of backdoor attacks on LLM-based agents. Intuitively, backdoor attacks often result in inconsistencies among the user's instruction, the agent's planning, and its execution. Drawing on this insight, ReAgent employs a two-level approach to detect potential backdoors. At the execution level, ReAgent verifies consistency between the agent's thoughts and actions; at the planning level, ReAgent leverages the agent's capability to reconstruct the instruction based on its thought trajectory, checking for consistency between the reconstructed instruction and the user's instruction. Extensive evaluation demonstrates ReAgent's effectiveness against various backdoor attacks across tasks. For instance, ReAgent reduces the attack success rate by up to 90\% in database operation tasks, outperforming existing defenses by large margins. This work reveals the potential of utilizing compromised agents themselves to mitigate backdoor risks. | [
{
"version": "v1",
"created": "Tue, 10 Jun 2025 01:45:56 GMT"
},
{
"version": "v2",
"created": "Wed, 11 Jun 2025 01:39:01 GMT"
}
] | 2025-06-12T00:00:00 | [
[
"Changjiang",
"Li",
""
],
[
"Jiacheng",
"Liang",
""
],
[
"Bochuan",
"Cao",
""
],
[
"Jinghui",
"Chen",
""
],
[
"Ting",
"Wang",
""
]
] |
2506.08356 | Shivang Chopra | Shivang Chopra, Gabriela Sanchez-Rodriguez, Lingchao Mao, Andrew J Feola, Jing Li, Zsolt Kira | MedMoE: Modality-Specialized Mixture of Experts for Medical Vision-Language Understanding | null | null | null | null | cs.CV | http://creativecommons.org/licenses/by/4.0/ | Different medical imaging modalities capture diagnostic information at varying spatial resolutions, from coarse global patterns to fine-grained localized structures. However, most existing vision-language frameworks in the medical domain apply a uniform strategy for local feature extraction, overlooking the modality-specific demands. In this work, we present MedMoE, a modular and extensible vision-language processing framework that dynamically adapts visual representation based on the diagnostic context. MedMoE incorporates a Mixture-of-Experts (MoE) module conditioned on the report type, which routes multi-scale image features through specialized expert branches trained to capture modality-specific visual semantics. These experts operate over feature pyramids derived from a Swin Transformer backbone, enabling spatially adaptive attention to clinically relevant regions. This framework produces localized visual representations aligned with textual descriptions, without requiring modality-specific supervision at inference. Empirical results on diverse medical benchmarks demonstrate that MedMoE improves alignment and retrieval performance across imaging modalities, underscoring the value of modality-specialized visual representations in clinical vision-language systems. | [
{
"version": "v1",
"created": "Tue, 10 Jun 2025 02:14:15 GMT"
},
{
"version": "v2",
"created": "Wed, 11 Jun 2025 12:15:03 GMT"
}
] | 2025-06-12T00:00:00 | [
[
"Chopra",
"Shivang",
""
],
[
"Sanchez-Rodriguez",
"Gabriela",
""
],
[
"Mao",
"Lingchao",
""
],
[
"Feola",
"Andrew J",
""
],
[
"Li",
"Jing",
""
],
[
"Kira",
"Zsolt",
""
]
] |
2506.08364 | Jash Parekh | Jash Rajesh Parekh, Pengcheng Jiang, Jiawei Han | CC-RAG: Structured Multi-Hop Reasoning via Theme-Based Causal Graphs | null | null | null | null | cs.CL | http://creativecommons.org/licenses/by/4.0/ | Understanding cause and effect relationships remains a formidable challenge for Large Language Models (LLMs), particularly in specialized domains where reasoning requires more than surface-level correlations. Retrieval-Augmented Generation (RAG) improves factual accuracy, but standard RAG pipelines treat evidence as flat context, lacking the structure required to model true causal dependencies. We introduce Causal-Chain RAG (CC-RAG), a novel approach that integrates zero-shot triple extraction and theme-aware graph chaining into the RAG pipeline, enabling structured multi-hop inference. Given a domain specific corpus, CC-RAG constructs a Directed Acyclic Graph (DAG) of <cause, relation, effect> triples and uses forward/backward chaining to guide structured answer generation. Experiments on two real-world domains: Bitcoin price fluctuations and Gaucher disease, show that CC-RAG outperforms standard RAG and zero-shot LLMs in chain similarity, information density, and lexical diversity. Both LLM-as-a-Judge and human evaluations consistently favor CC-RAG. Our results demonstrate that explicitly modeling causal structure enables LLMs to generate more accurate and interpretable responses, especially in specialized domains where flat retrieval fails. | [
{
"version": "v1",
"created": "Tue, 10 Jun 2025 02:22:32 GMT"
},
{
"version": "v2",
"created": "Wed, 11 Jun 2025 01:49:12 GMT"
}
] | 2025-06-12T00:00:00 | [
[
"Parekh",
"Jash Rajesh",
""
],
[
"Jiang",
"Pengcheng",
""
],
[
"Han",
"Jiawei",
""
]
] |
2506.08371 | Zikai Xiao | Zikai Xiao, Ziyang Wang, Wen Ma, Yan Zhang, Wei Shen, Yan Wang, Luqi Gong, and Zuozhu Liu | Mitigating Posterior Salience Attenuation in Long-Context LLMs with Positional Contrastive Decoding | null | null | null | null | cs.CL | http://creativecommons.org/publicdomain/zero/1.0/ | While Large Language Models (LLMs) support long contexts, they struggle with performance degradation within the context window. Current solutions incur prohibitive training costs, leaving statistical behaviors and cost-effective approaches underexplored. From the decoding perspective, we identify the Posterior Salience Attenuation (PSA) phenomenon, where the salience ratio correlates with long-text performance degradation. Notably, despite the attenuation, gold tokens still occupy high-ranking positions in the decoding space. Motivated by it, we propose the training-free Positional Contrastive Decoding (PCD) that contrasts the logits derived from long-aware attention with those from designed local-aware attention, enabling the model to focus on the gains introduced by large-scale short-to-long training. Through the analysis of long-term decay simulation, we demonstrate that PCD effectively alleviates attention score degradation. Experimental results show that PCD achieves state-of-the-art performance on long-context benchmarks. | [
{
"version": "v1",
"created": "Tue, 10 Jun 2025 02:35:26 GMT"
},
{
"version": "v2",
"created": "Wed, 11 Jun 2025 03:10:44 GMT"
}
] | 2025-06-12T00:00:00 | [
[
"Xiao",
"Zikai",
""
],
[
"Wang",
"Ziyang",
""
],
[
"Ma",
"Wen",
""
],
[
"Zhang",
"Yan",
""
],
[
"Shen",
"Wei",
""
],
[
"Wang",
"Yan",
""
],
[
"Gong",
"Luqi",
""
],
[
"Liu",
"Zuozhu",
""
]
] |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.