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 |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
2506.08381 | He Yang | He Yang, Fei Ren, Hai-Sui Yu, Xueyu Geng, Pei-Zhi Zhuang | TS-PIELM: Time-Stepping Physics-Informed Extreme Learning Machine Facilitates Soil Consolidation Analyses | null | null | null | null | physics.geo-ph cs.LG | http://creativecommons.org/licenses/by-nc-nd/4.0/ | Accuracy and efficiency of the conventional physics-informed neural network (PINN) need to be improved before it can be a competitive alternative for soil consolidation analyses. This paper aims to overcome these limitations by proposing a highly accurate and efficient physics-informed machine learning (PIML) approach, termed time-stepping physics-informed extreme learning machine (TS-PIELM). In the TS-PIELM framework the consolidation process is divided into numerous time intervals, which helps overcome the limitation of PIELM in solving differential equations with sharp gradients. To accelerate network training, the solution is approximated by a single-layer feedforward extreme learning machine (ELM), rather than using a fully connected neural network in PINN. The input layer weights of the ELM network are generated randomly and fixed during the training process. Subsequently, the output layer weights are directly computed by solving a system of linear equations, which significantly enhances the training efficiency compared to the time-consuming gradient descent method in PINN. Finally, the superior performance of TS-PIELM is demonstrated by solving three typical Terzaghi consolidation problems. Compared to PINN, results show that the computational efficiency and accuracy of the novel TS-PIELM framework are improved by more than 1000 times and 100 times for one-dimensional cases, respectively. This paper provides compelling evidence that PIML can be a powerful tool for computational geotechnics. | [
{
"version": "v1",
"created": "Tue, 10 Jun 2025 02:45:47 GMT"
},
{
"version": "v2",
"created": "Wed, 11 Jun 2025 03:02:30 GMT"
}
] | 2025-06-12T00:00:00 | [
[
"Yang",
"He",
""
],
[
"Ren",
"Fei",
""
],
[
"Yu",
"Hai-Sui",
""
],
[
"Geng",
"Xueyu",
""
],
[
"Zhuang",
"Pei-Zhi",
""
]
] |
2506.08392 | Zhenqi Jenny Wang | Zhenqi Jenny Wang | Multiple mixing and fractional cohomological equation | arXiv admin note: substantial text overlap with arXiv:math/0512192 by other authors | null | null | null | math.DS math.RT | http://creativecommons.org/licenses/by/4.0/ | We introduce the notion of the (multiple) fractional cohomological equation and, by studying its solutions, develop a novel framework to obtain the decay of matrix coefficients for partially hyperbolic algebraic actions. In particular, we show that mere partial H\"older regularity of \(L^2\) vectors is sufficient for exponential decay of matrix coefficients.
As an application, under the assumption of ergodicity, we obtain explicit and sharp exponential mixing rates of all orders for a large class of partially hyperbolic algebraic actions. Furthermore, we introduce the concept of irrational automorphisms on nilmanifolds and prove that these automorphisms exhibit super-exponential mixing of all orders, marking the first such example in the literature. | [
{
"version": "v1",
"created": "Tue, 10 Jun 2025 02:58:34 GMT"
},
{
"version": "v2",
"created": "Wed, 11 Jun 2025 03:43:04 GMT"
}
] | 2025-06-12T00:00:00 | [
[
"Wang",
"Zhenqi Jenny",
""
]
] |
2506.08399 | Jiachen Ma | Jiachen Ma, Zhanhui Zhou, Chao Yang, Chaochao Lu | SafeCoT: Improving VLM Safety with Minimal Reasoning | null | null | null | null | cs.AI cs.LG | http://creativecommons.org/licenses/by/4.0/ | Ensuring safe and appropriate responses from vision-language models (VLMs) remains a critical challenge, particularly in high-risk or ambiguous scenarios. We introduce SafeCoT, a lightweight, interpretable framework that leverages rule-based chain-of-thought (CoT) supervision to improve refusal behavior in VLMs. Unlike prior methods that rely on large-scale safety annotations or complex modeling, SafeCoT uses minimal supervision to help models reason about safety risks and make context-aware refusals. Experiments across multiple benchmarks show that SafeCoT significantly reduces overrefusal and enhances generalization, even with limited training data. Our approach offers a scalable solution for aligning VLMs with safety-critical objectives. | [
{
"version": "v1",
"created": "Tue, 10 Jun 2025 03:13:50 GMT"
},
{
"version": "v2",
"created": "Wed, 11 Jun 2025 06:57:37 GMT"
}
] | 2025-06-12T00:00:00 | [
[
"Ma",
"Jiachen",
""
],
[
"Zhou",
"Zhanhui",
""
],
[
"Yang",
"Chao",
""
],
[
"Lu",
"Chaochao",
""
]
] |
2506.08403 | Weiya Li | Weiya Li, Junjie Chen, Bei Li, Boyang Liu, Zichen Wen, Nuanqiao Shan, Xiaoqian Liu, Anping Liu, Huajie Liu, Hu Song, Linfeng Zhang | TACTIC: Translation Agents with Cognitive-Theoretic Interactive Collaboration | 20 pages, 4 figures, Under review. Code: https://github.com/weiyali126/TACTIC | null | null | null | cs.CL cs.AI | http://creativecommons.org/licenses/by/4.0/ | Machine translation has long been a central task in natural language processing. With the rapid advancement of large language models (LLMs), there has been remarkable progress in translation quality. However, fully realizing the translation potential of LLMs remains an open challenge. Recent studies have explored multi-agent systems to decompose complex translation tasks into collaborative subtasks, showing initial promise in enhancing translation quality through agent cooperation and specialization. Nevertheless, existing multi-agent translation frameworks largely neglect foundational insights from cognitive translation studies. These insights emphasize how human translators employ different cognitive strategies, such as balancing literal and free translation, refining expressions based on context, and iteratively evaluating outputs. To address this limitation, we propose a cognitively informed multi-agent framework called TACTIC, which stands for T ranslation A gents with Cognitive- T heoretic Interactive Collaboration. The framework comprises six functionally distinct agents that mirror key cognitive processes observed in human translation behavior. These include agents for drafting, refinement, evaluation, scoring, context reasoning, and external knowledge gathering. By simulating an interactive and theory-grounded translation workflow, TACTIC effectively leverages the full capacity of LLMs for high-quality translation. Experimental results on diverse language pairs from the FLORES-200 and WMT24 benchmarks show that our method consistently achieves state-of-the-art performance. Using DeepSeek-V3 as the base model, TACTIC surpasses GPT-4.1 by an average of +0.6 XCOMET and +1.18 COMETKIWI-23. Compared to DeepSeek-R1, it further improves by +0.84 XCOMET and +2.99 COMETKIWI-23. Code is available at https://github.com/weiyali126/TACTIC. | [
{
"version": "v1",
"created": "Tue, 10 Jun 2025 03:22:30 GMT"
},
{
"version": "v2",
"created": "Wed, 11 Jun 2025 15:57:34 GMT"
}
] | 2025-06-12T00:00:00 | [
[
"Li",
"Weiya",
""
],
[
"Chen",
"Junjie",
""
],
[
"Li",
"Bei",
""
],
[
"Liu",
"Boyang",
""
],
[
"Wen",
"Zichen",
""
],
[
"Shan",
"Nuanqiao",
""
],
[
"Liu",
"Xiaoqian",
""
],
[
"Liu",
"Anping",
""
],
[
"Liu",
"Huajie",
""
],
[
"Song",
"Hu",
""
],
[
"Zhang",
"Linfeng",
""
]
] |
2506.08422 | Ikkei Itoku Itoku | Ikkei Itoku, David Theil, Evelyn Eichelsdoerfer Uehara, Sreyoshi Bhaduri, Junnosuke Kuroda, Toshi Yumoto, Alex Gil, Natalie Perez, Rajesh Cherukuri, Naumaan Nayyar | Transforming Expert Knowledge into Scalable Ontology via Large Language Models | null | null | null | null | cs.AI | http://creativecommons.org/licenses/by/4.0/ | Having a unified, coherent taxonomy is essential for effective knowledge representation in domain-specific applications as diverse terminologies need to be mapped to underlying concepts. Traditional manual approaches to taxonomy alignment rely on expert review of concept pairs, but this becomes prohibitively expensive and time-consuming at scale, while subjective interpretations often lead to expert disagreements. Existing automated methods for taxonomy alignment have shown promise but face limitations in handling nuanced semantic relationships and maintaining consistency across different domains. These approaches often struggle with context-dependent concept mappings and lack transparent reasoning processes. We propose a novel framework that combines large language models (LLMs) with expert calibration and iterative prompt optimization to automate taxonomy alignment. Our method integrates expert-labeled examples, multi-stage prompt engineering, and human validation to guide LLMs in generating both taxonomy linkages and supporting rationales. In evaluating our framework on a domain-specific mapping task of concept essentiality, we achieved an F1-score of 0.97, substantially exceeding the human benchmark of 0.68. These results demonstrate the effectiveness of our approach in scaling taxonomy alignment while maintaining high-quality mappings and preserving expert oversight for ambiguous cases. | [
{
"version": "v1",
"created": "Tue, 10 Jun 2025 03:48:26 GMT"
},
{
"version": "v2",
"created": "Wed, 11 Jun 2025 03:16:55 GMT"
}
] | 2025-06-12T00:00:00 | [
[
"Itoku",
"Ikkei",
""
],
[
"Theil",
"David",
""
],
[
"Uehara",
"Evelyn Eichelsdoerfer",
""
],
[
"Bhaduri",
"Sreyoshi",
""
],
[
"Kuroda",
"Junnosuke",
""
],
[
"Yumoto",
"Toshi",
""
],
[
"Gil",
"Alex",
""
],
[
"Perez",
"Natalie",
""
],
[
"Cherukuri",
"Rajesh",
""
],
[
"Nayyar",
"Naumaan",
""
]
] |
2506.08424 | Yong Liang Goh | Yong Liang Goh, Zhiguang Cao, Yining Ma, Jianan Zhou, Mohammed Haroon Dupty, Wee Sun Lee | SHIELD: Multi-task Multi-distribution Vehicle Routing Solver with Sparsity and Hierarchy | Accepted in the 42nd International Conference of Machine Learning (ICML) | null | null | null | cs.AI | http://creativecommons.org/licenses/by-nc-nd/4.0/ | Recent advances toward foundation models for routing problems have shown great potential of a unified deep model for various VRP variants. However, they overlook the complex real-world customer distributions. In this work, we advance the Multi-Task VRP (MTVRP) setting to the more realistic yet challenging Multi-Task Multi-Distribution VRP (MTMDVRP) setting, and introduce SHIELD, a novel model that leverages both sparsity and hierarchy principles. Building on a deeper decoder architecture, we first incorporate the Mixture-of-Depths (MoD) technique to enforce sparsity. This improves both efficiency and generalization by allowing the model to dynamically select nodes to use or skip each decoder layer, providing the needed capacity to adaptively allocate computation for learning the task/distribution specific and shared representations. We also develop a context-based clustering layer that exploits the presence of hierarchical structures in the problems to produce better local representations. These two designs inductively bias the network to identify key features that are common across tasks and distributions, leading to significantly improved generalization on unseen ones. Our empirical results demonstrate the superiority of our approach over existing methods on 9 real-world maps with 16 VRP variants each. | [
{
"version": "v1",
"created": "Tue, 10 Jun 2025 03:55:14 GMT"
},
{
"version": "v2",
"created": "Wed, 11 Jun 2025 06:43:18 GMT"
}
] | 2025-06-12T00:00:00 | [
[
"Goh",
"Yong Liang",
""
],
[
"Cao",
"Zhiguang",
""
],
[
"Ma",
"Yining",
""
],
[
"Zhou",
"Jianan",
""
],
[
"Dupty",
"Mohammed Haroon",
""
],
[
"Lee",
"Wee Sun",
""
]
] |
2506.08431 | Neeraj Kumar Tomar | Neeraj Kumar Tomar, Fahed Zulfeqarr and M. C. Sharma | Softly $\pi g\hat{D}$-Normal Spaces | null | The Mathematics Education. Vol. LIX, No. 1, pp. 1-17, March 2025 | 10.5281/zenodo.15551191 | null | math.GN | http://creativecommons.org/licenses/by/4.0/ | The aim of this paper is to introduce a new class of softly normal called softly $\pi g\widehat{D}$ -normality by using $\pi g\widehat{D}$ -open sets and obtained several properties of such a space. We discuss many properties of this new space and we give some properties that connect this new spaces with some other topological spaces, also we present some examples and counter examples that show the relationships between softly $\pi g\widehat{D}$ -normal spaces and some other topological spaces, also we introduced the concept of $\pi g\widehat{D}$ -normal, almost $\pi g\widehat{D}$ -normal, quasi $\pi g\widehat{D}$ -normal, mildly $\pi g\widehat{D}$ -normal. The main result of this paper is that softly $\pi g\widehat{D}$ -normality is a topological property and it is a hereditary property with respect to closed domain subspaces. Moreover, we obtain some new characterizations and preservation theorems of softly $\pi g\widehat{D}$ -normal spaces. We insure that existence of utility for new results of softly $\pi g\widehat{D}$ -normality using separation axioms in topological spaces which is separate on a known separation axioms in topological spaces. | [
{
"version": "v1",
"created": "Tue, 10 Jun 2025 04:05:53 GMT"
}
] | 2025-06-12T00:00:00 | [
[
"Tomar",
"Neeraj Kumar",
""
],
[
"Zulfeqarr",
"Fahed",
""
],
[
"Sharma",
"M. C.",
""
]
] |
2506.08433 | Hern\'an Maina | Hern\'an Maina, Nicol\'as Wolovick, Luciana Benotti | Low-resource domain adaptation while minimizing energy and hardware resource consumption | A shorter version of this work was accepted as a two-page abstract for presentation at the Widening Natural Language Processing (WiNLP) 2023 Workshop. That version was not publicly released, and this is the first public version of the work | null | null | null | cs.CL cs.DC cs.LG | http://creativecommons.org/licenses/by-nc-sa/4.0/ | Training Large Language Models (LLMs) is costly in terms of energy, hardware, and annotated data, often resulting in a positionality rooted in predominant cultures and values (Santy et al., 2023). Domain adaptation has emerged as a promising strategy to better align models with diverse cultural and value contexts (Hershcovich et al., 2022), but its computational cost remains a significant barrier, particularly for research groups lacking access to large-scale infrastructure. In this paper, we evaluate how the use of different numerical precision formats and data parallelization strategies impacts both training speed (as a proxy to energy and hardware consumption) and model accuracy, with the goal of facilitating domain adaptation in low-resource environments. Our findings are relevant to any setting where energy efficiency, accessibility, or limited hardware availability are key concerns. | [
{
"version": "v1",
"created": "Tue, 10 Jun 2025 04:14:19 GMT"
},
{
"version": "v2",
"created": "Wed, 11 Jun 2025 03:45:24 GMT"
}
] | 2025-06-12T00:00:00 | [
[
"Maina",
"Hernán",
""
],
[
"Wolovick",
"Nicolás",
""
],
[
"Benotti",
"Luciana",
""
]
] |
2506.08440 | Zengjue Chen | Zengjue Chen, Runliang Niu, He Kong, Qi Wang | TGRPO :Fine-tuning Vision-Language-Action Model via Trajectory-wise Group Relative Policy Optimization | null | null | null | null | cs.RO cs.AI | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Recent advances in Vision-Language-Action (VLA) model have demonstrated strong generalization capabilities across diverse scenes, tasks, and robotic platforms when pretrained at large-scale datasets. However, these models still require task-specific fine-tuning in novel environments, a process that relies almost exclusively on supervised fine-tuning (SFT) using static trajectory datasets. Such approaches neither allow robot to interact with environment nor do they leverage feedback from live execution. Also, their success is critically dependent on the size and quality of the collected trajectories. Reinforcement learning (RL) offers a promising alternative by enabling closed-loop interaction and aligning learned policies directly with task objectives. In this work, we draw inspiration from the ideas of GRPO and propose the Trajectory-wise Group Relative Policy Optimization (TGRPO) method. By fusing step-level and trajectory-level advantage signals, this method improves GRPO's group-level advantage estimation, thereby making the algorithm more suitable for online reinforcement learning training of VLA. Experimental results on ten manipulation tasks from the libero-object benchmark demonstrate that TGRPO consistently outperforms various baseline methods, capable of generating more robust and efficient policies across multiple tested scenarios. Our source codes are available at: https://github.com/hahans/TGRPO | [
{
"version": "v1",
"created": "Tue, 10 Jun 2025 04:27:49 GMT"
},
{
"version": "v2",
"created": "Wed, 11 Jun 2025 04:42:44 GMT"
}
] | 2025-06-12T00:00:00 | [
[
"Chen",
"Zengjue",
""
],
[
"Niu",
"Runliang",
""
],
[
"Kong",
"He",
""
],
[
"Wang",
"Qi",
""
]
] |
2506.08452 | Tianya Xia | Tianya Xia, Li Lin Yang | Reconstructing Laurent expansion of rational functions using p-adic numbers | null | null | null | null | hep-th hep-ph | http://creativecommons.org/licenses/by/4.0/ | We propose a novel method for reconstructing Laurent expansion of rational functions using $p$-adic numbers. By evaluating the rational functions in $p$-adic fields rather than finite fields, it is possible to probe the expansion coefficients simultaneously, enabling their reconstruction from a single set of evaluations. Compared with the reconstruction of the full expression, constructing the Laurent expansion to the first few orders significantly reduces the required computational resources. Our method can handle expansions with respect to more than one variables simultaneously. Among possible applications, we anticipate that our method can be used to simplify the integration-by-parts reduction of Feynman integrals in cutting-edge calculations. | [
{
"version": "v1",
"created": "Tue, 10 Jun 2025 05:04:47 GMT"
}
] | 2025-06-12T00:00:00 | [
[
"Xia",
"Tianya",
""
],
[
"Yang",
"Li Lin",
""
]
] |
2506.08454 | Kevin Zelaya | Matthew Markowitz and Mohammad-Ali Miri and Alexey Ovchinnikov and Kevin Zelaya | Universality of Photonic Interlacing Architectures for Learning Discrete Linear Unitaries | null | null | null | null | quant-ph math-ph math.MP physics.optics | http://creativecommons.org/licenses/by-nc-nd/4.0/ | Recent investigations suggest that the discrete linear unitary group $U(N)$ can be represented by interlacing a finite sequence of diagonal phase operations with an intervening unitary operator. However, despite rigorous numerical justifications, no formal proof has been provided. Here, we show that elements of $U(N)$ can be decomposed into a sequence of $N$-parameter phases alternating with $1$-parameter propagators of a lattice Hamiltonian. The proof is based on building a Lie group by alternating these two operators and showing its completeness to represent $U(N)$ for a finite number of layers, which is numerically found to be exactly $N$. This architecture can be implemented using elementary optical components and can successfully reconstruct arbitrary unitary matrices. We propose example devices such as optical logic gates, which perform logic gate operations using a single-layer lossless and passive optical circuit design. | [
{
"version": "v1",
"created": "Tue, 10 Jun 2025 05:18:47 GMT"
},
{
"version": "v2",
"created": "Wed, 11 Jun 2025 15:31:13 GMT"
}
] | 2025-06-12T00:00:00 | [
[
"Markowitz",
"Matthew",
""
],
[
"Miri",
"Mohammad-Ali",
""
],
[
"Ovchinnikov",
"Alexey",
""
],
[
"Zelaya",
"Kevin",
""
]
] |
2506.08465 | Michael V. Klibanov | Michael V. Klibanov and Kevin McGoff and Trung Truong | Forecasting Public Sentiments via Mean Field Games | 26 pages | null | null | null | math.NA cs.NA | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | A mathematical model for forecasting of public sentiments via the Mean Field Games theory is proposed. A numerical method is developed. This is a version of the so-called convexification method. Convergence analysis demonstrates the global convergence of this method. Convergence rate is established. Numerical experiments demonstrate both an accurate performance of the convexification technique and some promising features of this approach. | [
{
"version": "v1",
"created": "Tue, 10 Jun 2025 05:38:27 GMT"
},
{
"version": "v2",
"created": "Wed, 11 Jun 2025 03:45:27 GMT"
}
] | 2025-06-12T00:00:00 | [
[
"Klibanov",
"Michael V.",
""
],
[
"McGoff",
"Kevin",
""
],
[
"Truong",
"Trung",
""
]
] |
2506.08473 | Shuo Yang | Shuo Yang and Qihui Zhang and Yuyang Liu and Yue Huang and Xiaojun Jia and Kunpeng Ning and Jiayu Yao and Jigang Wang and Hailiang Dai and Yibing Song and Li Yuan | AsFT: Anchoring Safety During LLM Fine-Tuning Within Narrow Safety Basin | null | null | null | null | cs.LG | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Large language models (LLMs) are vulnerable to safety risks during fine-tuning, where small amounts of malicious or harmless data can compromise safeguards. In this paper, building on the concept of alignment direction -- defined by the weight difference between aligned and unaligned models -- we observe that perturbations along this direction preserve model safety. In contrast, perturbations along directions orthogonal to this alignment are strongly linked to harmful direction perturbations, rapidly degrading safety and framing the parameter space as a narrow safety basin. Based on this insight, we propose a methodology for safety fine-tuning called AsFT (Anchoring Safety in Fine-Tuning), which integrates a regularization term into the training objective. This term uses the alignment direction as an anchor to suppress updates in harmful directions, ensuring that fine-tuning is constrained within the narrow safety basin. Extensive experiments on multiple datasets show that AsFT outperforms Safe LoRA, reducing harmful behavior by 7.60 percent, improving model performance by 3.44 percent, and maintaining robust performance across various experimental settings. Code is available at https://github.com/PKU-YuanGroup/AsFT | [
{
"version": "v1",
"created": "Tue, 10 Jun 2025 05:59:48 GMT"
},
{
"version": "v2",
"created": "Wed, 11 Jun 2025 02:43:10 GMT"
}
] | 2025-06-12T00:00:00 | [
[
"Yang",
"Shuo",
""
],
[
"Zhang",
"Qihui",
""
],
[
"Liu",
"Yuyang",
""
],
[
"Huang",
"Yue",
""
],
[
"Jia",
"Xiaojun",
""
],
[
"Ning",
"Kunpeng",
""
],
[
"Yao",
"Jiayu",
""
],
[
"Wang",
"Jigang",
""
],
[
"Dai",
"Hailiang",
""
],
[
"Song",
"Yibing",
""
],
[
"Yuan",
"Li",
""
]
] |
2506.08478 | Animesh Bhandari | Animesh Bhandari | Short note on phase retrievable weaving fusion frames | arXiv admin note: text overlap with arXiv:2409.01288 | null | null | null | math.FA | http://creativecommons.org/licenses/by/4.0/ | Fusion frames are extensively studied due to their effectiveness in recovering signals from large-scale data. They are applicable in distributed processing, wireless sensor networks, and packet encoding systems due to their robustness and redundancy. Motivated by the foundational work of Bemrose et al.\cite{Be16} and Balan\cite{Ba13}, this paper investigates the theoretical properties and characterizations of phase retrievable weaving fusion frames. These frames offer enhanced redundancy and stability in signal reconstruction. We present key results that deepen the understanding of their structure and behaviour. Lastly, an application involving probabilistic erasure is explored to demonstrate their practical utility. | [
{
"version": "v1",
"created": "Tue, 10 Jun 2025 06:10:47 GMT"
},
{
"version": "v2",
"created": "Wed, 11 Jun 2025 06:17:33 GMT"
}
] | 2025-06-12T00:00:00 | [
[
"Bhandari",
"Animesh",
""
]
] |
2506.08481 | Enrico Bernardi | Enrico Bernardi, Tatsuo Nishitani | Notes on tangent bicharacteristics and ill-posedness of the Cauchy problem | null | null | null | null | math.AP math.DS | http://creativecommons.org/licenses/by/4.0/ | We exhibit a family of second-order hyperbolic differential operators presenting spectral transition of the Hamilton map. As a consequence we prove that the Cauchy problem is not locally solvable at the origin in Gevrey classes of order greater than some fixed value. The main feature of these operators is that they may all have bicharacteristics tangent to the double manifold. | [
{
"version": "v1",
"created": "Tue, 10 Jun 2025 06:12:07 GMT"
},
{
"version": "v2",
"created": "Wed, 11 Jun 2025 05:16:36 GMT"
}
] | 2025-06-12T00:00:00 | [
[
"Bernardi",
"Enrico",
""
],
[
"Nishitani",
"Tatsuo",
""
]
] |
2506.08499 | Ramon Brasser | R. Brasser | Impact chronology of leftover planetesimals | Submitted to Astronomy and Astrophysics | null | null | null | astro-ph.EP | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | After the formation of the Moon the terrestrial planets were pummelled by impacts from planetesimals left over from terrestrial planet formation. This work attempts to reproduce the impact rates set by modern crater chronologies using leftover planetesimals from three different dynamical models of terrestrial planet formation. I ran dynamical simulations for 1 billion years using leftover planetesimals from the Grand Tack, Depleted Disc and Implantation models of terrestrial planet formation with the GENGA N-body integrator. I fit the cumulative impacts on the Earth and Mars using a function that is a sum of exponentials with different weighing factors and e-folding times. Most fits require three or four terms. The fitted timescales cluster around t1=10 Myr, t2=35 Myr, t3=100 Myr and t4>200 Myr. I attribute them to dynamical losses of planetesimals through different mechanisms: high-eccentricity Earth crossers and the nu6 secular resonance, Earth crossers, Mars crossers, and objects leaking on to Mars crossing orbits from beyond Mars. I place a constraint on the initial population using the known Archean terrestrial spherule beds, and I conclude that the Archean impacts were mostly created by leftover planetesimals. The inferred mass in leftover planetesimals at the time of the Moon's formation was about 0.015 Earth masses. The third time constant is comparable to that of modern crater chronologies. As such, the crater chronologies are indicative of impacts by an ancient population of Mars crossers. The initial perihelion distribution of the leftovers is a major factor in setting the rate of decline: to reproduce the current crater chronologies the number of Earth crossers at the time of the Moon's formation had to be at most half of the Mars crossers. These results together place constraints on dynamical models of terrestrial planet formation. | [
{
"version": "v1",
"created": "Tue, 10 Jun 2025 06:52:24 GMT"
},
{
"version": "v2",
"created": "Wed, 11 Jun 2025 05:25:50 GMT"
}
] | 2025-06-12T00:00:00 | [
[
"Brasser",
"R.",
""
]
] |
2506.08524 | Weiguo Wang | Weiguo Wang, Andy Nie, Wenrui Zhou, Yi Kai, Chengchen Hu | Teaching Physical Awareness to LLMs through Sounds | ICML 2025 | null | null | null | cs.SD cs.AI cs.MM cs.RO eess.AS | http://creativecommons.org/licenses/by/4.0/ | Large Language Models (LLMs) have shown remarkable capabilities in text and multimodal processing, yet they fundamentally lack physical awareness--understanding of real-world physical phenomena. In this work, we present ACORN, a framework that teaches LLMs physical awareness through sound, focusing on fundamental physical phenomena like the Doppler effect, multipath effect, and spatial relationships. To overcome data scarcity, ACORN introduce a physics-based simulator combining real-world sound sources with controlled physical channels to generate diverse training data. Using this simulator, we build AQA-PHY, a comprehensive Audio Question-Answer dataset, and propose an audio encoder that processes both magnitude and phase information. By connecting our audio encoder to state-of-the-art LLMs, we demonstrate reasonable results in both simulated and real-world tasks, such as line-of-sight detection, Doppler effect estimation, and Direction-of-Arrival estimation, paving the way for enabling LLMs to understand physical world. | [
{
"version": "v1",
"created": "Tue, 10 Jun 2025 07:42:56 GMT"
},
{
"version": "v2",
"created": "Wed, 11 Jun 2025 05:18:01 GMT"
}
] | 2025-06-12T00:00:00 | [
[
"Wang",
"Weiguo",
""
],
[
"Nie",
"Andy",
""
],
[
"Zhou",
"Wenrui",
""
],
[
"Kai",
"Yi",
""
],
[
"Hu",
"Chengchen",
""
]
] |
2506.08536 | Nicol\`o Maccaferri | Tlek Tapani, Jonas M. Pettersson, Nils Henriksson, Erik Z\"all, Nils V. Hauff, Lakshmi Das, Gianluca Balestra, Massimo Cuscun\`a, Aitor De Andr\'es, Tommaso Giovannini, Denis Garoli, and Nicol\`o Maccaferri | Ultrafast interband transitions in nanoporous gold metamaterial | null | null | null | null | cond-mat.mes-hall cond-mat.mtrl-sci | http://creativecommons.org/licenses/by/4.0/ | Nanoporous metals have emerged as promising functional architectures due to their tunable optical and electronic properties, high surface areas, and versatile use in real-life applications such as sensing, catalysis, and biomedicine. While the optical and morphological properties of nanoporous metals have been extensively studied, their electronic properties at ultrafast timescales remain largely unexplored. Here, we study the transient response of a nanoporous gold metamaterial and compare it with the ultrafast dynamics of a continuous gold film. We unravel that the nanoporous sample supports lower energy interband transitions, due to a much higher electron temperature in the nanoporous material, which causes an enhanced redistribution of electron density around the Fermi level. The experimental results are consistent with the two-temperature model, which highlights the role of nanoscale porosity in enabling the more efficient generation of hot carriers, thus allowing lower energy photons to induce interband transitions. Our findings demonstrate that nanoporosity affects fundamental ultrafast electronic processes and introduces this platform as temporal metamaterial allowing the emergence of tunable electronic properties not supported by the bulk counterpart. Furthermore, we present new insights into ultrafast electronic properties of nanoporous metals, which can impact several areas, from photochemistry and catalysis to energy harvesting and opto-electronics. | [
{
"version": "v1",
"created": "Tue, 10 Jun 2025 07:59:38 GMT"
},
{
"version": "v2",
"created": "Wed, 11 Jun 2025 14:52:14 GMT"
}
] | 2025-06-12T00:00:00 | [
[
"Tapani",
"Tlek",
""
],
[
"Pettersson",
"Jonas M.",
""
],
[
"Henriksson",
"Nils",
""
],
[
"Zäll",
"Erik",
""
],
[
"Hauff",
"Nils V.",
""
],
[
"Das",
"Lakshmi",
""
],
[
"Balestra",
"Gianluca",
""
],
[
"Cuscunà",
"Massimo",
""
],
[
"De Andrés",
"Aitor",
""
],
[
"Giovannini",
"Tommaso",
""
],
[
"Garoli",
"Denis",
""
],
[
"Maccaferri",
"Nicolò",
""
]
] |
2506.08551 | Ting Wang | Panlong Wu, Ting Wang, Yifei Zhong, Haoqi Zhang, Zitong Wang and Fangxin Wang | DeepForm: Reasoning Large Language Model for Communication System Formulation | null | null | null | null | cs.LG | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Communication system formulation is critical for advancing 6G and future wireless technologies, yet it remains a complex, expertise-intensive task. While Large Language Models (LLMs) offer potential, existing general-purpose models often lack the specialized domain knowledge, nuanced reasoning capabilities, and access to high-quality, domain-specific training data required for adapting a general LLM into an LLM specially for communication system formulation. To bridge this gap, we introduce DeepForm, the first reasoning LLM specially for automated communication system formulation. We propose the world-first large-scale, open-source dataset meticulously curated for this domain called Communication System Formulation Reasoning Corpus (CSFRC). Our framework employs a two-stage training strategy: first, Supervised Fine-Tuning (SFT) with Chain-of-Thought (CoT) data to distill domain knowledge; second, a novel rule-based Reinforcement Learning (RL) algorithm, C-ReMax based on ReMax, to cultivate advanced modeling capabilities and elicit sophisticated reasoning patterns like self-correction and verification. Extensive experiments demonstrate that our model achieves state-of-the-art performance, significantly outperforming larger proprietary LLMs on diverse senerios. We will release related resources to foster further research in this area after the paper is accepted. | [
{
"version": "v1",
"created": "Tue, 10 Jun 2025 08:16:24 GMT"
},
{
"version": "v2",
"created": "Wed, 11 Jun 2025 05:56:27 GMT"
}
] | 2025-06-12T00:00:00 | [
[
"Wu",
"Panlong",
""
],
[
"Wang",
"Ting",
""
],
[
"Zhong",
"Yifei",
""
],
[
"Zhang",
"Haoqi",
""
],
[
"Wang",
"Zitong",
""
],
[
"Wang",
"Fangxin",
""
]
] |
2506.08558 | William De Vazelhes | William de Vazelhes, Xiao-Tong Yuan, Bin Gu | Optimization over Sparse Support-Preserving Sets: Two-Step Projection with Global Optimality Guarantees | Accepted for publication at ICML 2025 | null | null | null | math.OC cs.LG | http://creativecommons.org/licenses/by/4.0/ | In sparse optimization, enforcing hard constraints using the $\ell_0$ pseudo-norm offers advantages like controlled sparsity compared to convex relaxations. However, many real-world applications demand not only sparsity constraints but also some extra constraints. While prior algorithms have been developed to address this complex scenario with mixed combinatorial and convex constraints, they typically require the closed form projection onto the mixed constraints which might not exist, and/or only provide local guarantees of convergence which is different from the global guarantees commonly sought in sparse optimization. To fill this gap, in this paper, we study the problem of sparse optimization with extra support-preserving constraints commonly encountered in the literature. We present a new variant of iterative hard-thresholding algorithm equipped with a two-step consecutive projection operator customized for these mixed constraints, serving as a simple alternative to the Euclidean projection onto the mixed constraint. By introducing a novel trade-off between sparsity relaxation and sub-optimality, we provide global guarantees in objective value for the output of our algorithm, in the deterministic, stochastic, and zeroth-order settings, under the conventional restricted strong-convexity/smoothness assumptions. As a fundamental contribution in proof techniques, we develop a novel extension of the classic three-point lemma to the considered two-step non-convex projection operator, which allows us to analyze the convergence in objective value in an elegant way that has not been possible with existing techniques. In the zeroth-order case, such technique also improves upon the state-of-the-art result from de Vazelhes et. al. (2022), even in the case without additional constraints, by allowing us to remove a non-vanishing system error present in their work. | [
{
"version": "v1",
"created": "Tue, 10 Jun 2025 08:27:01 GMT"
},
{
"version": "v2",
"created": "Wed, 11 Jun 2025 06:46:49 GMT"
}
] | 2025-06-12T00:00:00 | [
[
"de Vazelhes",
"William",
""
],
[
"Yuan",
"Xiao-Tong",
""
],
[
"Gu",
"Bin",
""
]
] |
2506.08561 | Hao Wu | Hao Wu, Haijun Wang, Shangwang Li, Yin Wu, Ming Fan, Yitao Zhao, Ting Liu | Detecting State Manipulation Vulnerabilities in Smart Contracts Using LLM and Static Analysis | null | null | null | null | cs.SE | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | An increasing number of DeFi protocols are gaining popularity, facilitating transactions among multiple anonymous users. State Manipulation is one of the notorious attacks in DeFi smart contracts, with price variable being the most commonly exploited state variable-attackers manipulate token prices to gain illicit profits. In this paper, we propose PriceSleuth, a novel method that leverages the Large Language Model (LLM) and static analysis to detect Price Manipulation (PM) attacks proactively. PriceSleuth firstly identifies core logic function related to price calculation in DeFi contracts. Then it guides LLM to locate the price calculation code statements. Secondly, PriceSleuth performs backward dependency analysis of price variables, instructing LLM in detecting potential price manipulation. Finally, PriceSleuth utilizes propagation analysis of price variables to assist LLM in detecting whether these variables are maliciously exploited. We presented preliminary experimental results to substantiate the effectiveness of PriceSleuth . And we outline future research directions for PriceSleuth. | [
{
"version": "v1",
"created": "Tue, 10 Jun 2025 08:31:44 GMT"
},
{
"version": "v2",
"created": "Wed, 11 Jun 2025 02:29:28 GMT"
}
] | 2025-06-12T00:00:00 | [
[
"Wu",
"Hao",
""
],
[
"Wang",
"Haijun",
""
],
[
"Li",
"Shangwang",
""
],
[
"Wu",
"Yin",
""
],
[
"Fan",
"Ming",
""
],
[
"Zhao",
"Yitao",
""
],
[
"Liu",
"Ting",
""
]
] |
2506.08563 | Siyuan Yang | Siyuan Yang, Cheng Song, Zhilu Lai, Wenjia Wang | KP-PINNs: Kernel Packet Accelerated Physics Informed Neural Networks | Accepted to IJCAI 2025 | null | null | null | cs.CE cs.AI math-ph math.MP physics.comp-ph | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Differential equations are involved in modeling many engineering problems. Many efforts have been devoted to solving differential equations. Due to the flexibility of neural networks, Physics Informed Neural Networks (PINNs) have recently been proposed to solve complex differential equations and have demonstrated superior performance in many applications. While the L2 loss function is usually a default choice in PINNs, it has been shown that the corresponding numerical solution is incorrect and unstable for some complex equations. In this work, we propose a new PINNs framework named Kernel Packet accelerated PINNs (KP-PINNs), which gives a new expression of the loss function using the reproducing kernel Hilbert space (RKHS) norm and uses the Kernel Packet (KP) method to accelerate the computation. Theoretical results show that KP-PINNs can be stable across various differential equations. Numerical experiments illustrate that KP-PINNs can solve differential equations effectively and efficiently. This framework provides a promising direction for improving the stability and accuracy of PINNs-based solvers in scientific computing. | [
{
"version": "v1",
"created": "Tue, 10 Jun 2025 08:33:34 GMT"
},
{
"version": "v2",
"created": "Wed, 11 Jun 2025 08:00:24 GMT"
}
] | 2025-06-12T00:00:00 | [
[
"Yang",
"Siyuan",
""
],
[
"Song",
"Cheng",
""
],
[
"Lai",
"Zhilu",
""
],
[
"Wang",
"Wenjia",
""
]
] |
2506.08570 | Or Tal | Or Tal and Felix Kreuk and Yossi Adi | Auto-Regressive vs Flow-Matching: a Comparative Study of Modeling Paradigms for Text-to-Music Generation | null | null | null | null | cs.SD cs.AI cs.LG eess.AS | http://creativecommons.org/licenses/by-nc-sa/4.0/ | Recent progress in text-to-music generation has enabled models to synthesize high-quality musical segments, full compositions, and even respond to fine-grained control signals, e.g. chord progressions. State-of-the-art (SOTA) systems differ significantly across many dimensions, such as training datasets, modeling paradigms, and architectural choices. This diversity complicates efforts to evaluate models fairly and pinpoint which design choices most influence performance. While factors like data and architecture are important, in this study we focus exclusively on the modeling paradigm. We conduct a systematic empirical analysis to isolate its effects, offering insights into associated trade-offs and emergent behaviors that can guide future text-to-music generation systems. Specifically, we compare the two arguably most common modeling paradigms: Auto-Regressive decoding and Conditional Flow-Matching. We conduct a controlled comparison by training all models from scratch using identical datasets, training configurations, and similar backbone architectures. Performance is evaluated across multiple axes, including generation quality, robustness to inference configurations, scalability, adherence to both textual and temporally aligned conditioning, and editing capabilities in the form of audio inpainting. This comparative study sheds light on distinct strengths and limitations of each paradigm, providing actionable insights that can inform future architectural and training decisions in the evolving landscape of text-to-music generation. Audio sampled examples are available at: https://huggingface.co/spaces/ortal1602/ARvsFM | [
{
"version": "v1",
"created": "Tue, 10 Jun 2025 08:37:45 GMT"
},
{
"version": "v2",
"created": "Wed, 11 Jun 2025 13:09:05 GMT"
}
] | 2025-06-12T00:00:00 | [
[
"Tal",
"Or",
""
],
[
"Kreuk",
"Felix",
""
],
[
"Adi",
"Yossi",
""
]
] |
2506.08574 | Luigi Fiorillo | Alvise Dei Rossi, Matteo Metaldi, Michal Bechny, Irina Filchenko, Julia van der Meer, Markus H. Schmidt, Claudio L.A. Bassetti, Athina Tzovara, Francesca D. Faraci and Luigi Fiorillo | SLEEPYLAND: trust begins with fair evaluation of automatic sleep staging models | 41 pages, 4 Figures, 7 Tables | null | null | null | cs.LG | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Despite advances in deep learning for automatic sleep staging, clinical adoption remains limited due to challenges in fair model evaluation, generalization across diverse datasets, model bias, and variability in human annotations. We present SLEEPYLAND, an open-source sleep staging evaluation framework designed to address these barriers. It includes more than 220'000 hours in-domain (ID) sleep recordings, and more than 84'000 hours out-of-domain (OOD) sleep recordings, spanning a broad range of ages, sleep-wake disorders, and hardware setups. We release pre-trained models based on high-performing SoA architectures and evaluate them under standardized conditions across single- and multi-channel EEG/EOG configurations. We introduce SOMNUS, an ensemble combining models across architectures and channel setups via soft voting. SOMNUS achieves robust performance across twenty-four different datasets, with macro-F1 scores between 68.7% and 87.2%, outperforming individual models in 94.9% of cases. Notably, SOMNUS surpasses previous SoA methods, even including cases where compared models were trained ID while SOMNUS treated the same data as OOD. Using a subset of the BSWR (N=6'633), we quantify model biases linked to age, gender, AHI, and PLMI, showing that while ensemble improves robustness, no model architecture consistently minimizes bias in performance and clinical markers estimation. In evaluations on OOD multi-annotated datasets (DOD-H, DOD-O), SOMNUS exceeds the best human scorer, i.e., MF1 85.2% vs 80.8% on DOD-H, and 80.2% vs 75.9% on DOD-O, better reproducing the scorer consensus than any individual expert (k = 0.89/0.85 and ACS = 0.95/0.94 for healthy/OSA cohorts). Finally, we introduce ensemble disagreement metrics - entropy and inter-model divergence based - predicting regions of scorer disagreement with ROC AUCs up to 0.828, offering a data-driven proxy for human uncertainty. | [
{
"version": "v1",
"created": "Tue, 10 Jun 2025 08:46:19 GMT"
},
{
"version": "v2",
"created": "Wed, 11 Jun 2025 13:12:29 GMT"
}
] | 2025-06-12T00:00:00 | [
[
"Rossi",
"Alvise Dei",
""
],
[
"Metaldi",
"Matteo",
""
],
[
"Bechny",
"Michal",
""
],
[
"Filchenko",
"Irina",
""
],
[
"van der Meer",
"Julia",
""
],
[
"Schmidt",
"Markus H.",
""
],
[
"Bassetti",
"Claudio L. A.",
""
],
[
"Tzovara",
"Athina",
""
],
[
"Faraci",
"Francesca D.",
""
],
[
"Fiorillo",
"Luigi",
""
]
] |
2506.08579 | Guiyang Luo | Guiyang Luo, Jinglin Li, Qixun Zhang, Zhiyong Feng, Quan Yuan, Yijing Lin, Hui Zhang, Nan Cheng, Ping Zhang | Toward Low-Altitude Airspace Management and UAV Operations: Requirements, Architecture and Enabling Technologies | null | null | null | null | eess.SY cs.SY | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | The low-altitude economy (LAE) is rapidly advancing toward intelligence, connectivity, and coordination, bringing new challenges in dynamic airspace management, unmanned aerial vehicle (UAV) operation, and security management. Existing systems remain fragmented and lack effective coordination. To bridge these gaps, we propose UTICN (Ubiquitous and Trusted Intelligent Cellular-native Network) for LAE, a unified cellular-native architecture that integrates multi-domain sensing, high-precision positioning, intelligent aircraft-to-everything communication, dynamic airspace management, and UAV operational services. UTICN introduces key technologies such as integrated sensing and communication (ISAC), passive and active positioning, intelligent machine communication, swarm coordination, and control-data decoupled management frameworks. We demonstrate UTICN's feasibility through two use cases, i.e., a city-level LAE management platform and a multi-frequency collaborative ISAC system. This work provides a fundamental reference for building a unified operational foundation and airspace management architecture for the LAE. | [
{
"version": "v1",
"created": "Tue, 10 Jun 2025 08:51:16 GMT"
},
{
"version": "v2",
"created": "Wed, 11 Jun 2025 03:32:54 GMT"
}
] | 2025-06-12T00:00:00 | [
[
"Luo",
"Guiyang",
""
],
[
"Li",
"Jinglin",
""
],
[
"Zhang",
"Qixun",
""
],
[
"Feng",
"Zhiyong",
""
],
[
"Yuan",
"Quan",
""
],
[
"Lin",
"Yijing",
""
],
[
"Zhang",
"Hui",
""
],
[
"Cheng",
"Nan",
""
],
[
"Zhang",
"Ping",
""
]
] |
2506.08581 | Fabian Pe\~na | Fabian C. Pe\~na, Steffen Herbold | Evaluating the Performance and Efficiency of Sentence-BERT for Code Comment Classification | 4th Intl. Workshop on NL-based Software Engineering (NLBSE) | 2025 IEEE/ACM International Workshop on Natural Language-Based Software Engineering (NLBSE) | null | null | cs.SE | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | This work evaluates Sentence-BERT for a multi-label code comment classification task seeking to maximize the classification performance while controlling efficiency constraints during inference. Using a dataset of 13,216 labeled comment sentences, Sentence-BERT models are fine-tuned and combined with different classification heads to recognize comment types. While larger models outperform smaller ones in terms of F1, the latter offer outstanding efficiency, both in runtime and GFLOPS. As result, a balance between a reasonable F1 improvement (+0.0346) and a minimal efficiency degradation (+1.4x in runtime and +2.1x in GFLOPS) is reached. | [
{
"version": "v1",
"created": "Tue, 10 Jun 2025 08:51:21 GMT"
}
] | 2025-06-12T00:00:00 | [
[
"Peña",
"Fabian C.",
""
],
[
"Herbold",
"Steffen",
""
]
] |
2506.08582 | Laura Freijeiro-Gonz\'alez | Laura Freijeiro-Gonz\'alez, Manuel Febrero-Bande and Wenceslao Gonz\'alez-Manteiga | Guidelines for LASSO and derivatives use under different dependence and scale structures | null | null | null | null | stat.ME | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | In a multivariate linear regression model with $p>1$ covariates, implementation of penalization techniques often implies a preliminary univariate standardization step. Although this prevents scale effects on the covariates selection procedure, possible dependence structures can be disrupted, leading to wrong results. This is particularly challenging in high-dimensional settings where $p \geq n$. In this paper, we analyze the standardization effect on the LASSO for different dependence-scales contexts by means of an extensive simulation study. Two distinct objectives are pursued: adequate covariate selection and proper predictive capability. Additionally, its behavior is compared with the one of some well-known or innovative competitors. This comparison is also extended to three real datasets facing different dependence-scales patterns. Eventually, we conclude with discussion and guidelines on the most suitable methodology for each case in terms of covariates selection or prediction. | [
{
"version": "v1",
"created": "Tue, 10 Jun 2025 08:52:20 GMT"
},
{
"version": "v2",
"created": "Wed, 11 Jun 2025 08:50:35 GMT"
}
] | 2025-06-12T00:00:00 | [
[
"Freijeiro-González",
"Laura",
""
],
[
"Febrero-Bande",
"Manuel",
""
],
[
"González-Manteiga",
"Wenceslao",
""
]
] |
2506.08595 | Harikumar Chandrasekharan Mr | Harikumar K Chandrasekharan and Ross Donaldson | High-throughput polarization-independent spatial mode shaping with mode-selective photonic lanterns | 5 pages, 3 figures | null | null | null | physics.optics | http://creativecommons.org/licenses/by-nc-nd/4.0/ | Coherent interference in multimode photonic systems offers a transformative approach for scalable, high-fidelity optical control, enabling precise beam shaping, efficient power delivery, and advanced signal processing through single-mode performance. Current techniques are often constrained to a single polarization state of light, system complexity, and the fabrication challenges associated with precise waveguide structures. Here, we present a high-efficiency coherent combination scheme employing a commercial mode-selective photonic lantern supporting three spatial channels, achieving Gaussian mode interference at multiple locations along the multimode core with up to 100% mode conversion efficiencies. By integrating a Faraday reflecting fibre component and piezo-phase shifters with the photonic lantern channels, we demonstrate stable and adaptive mode manipulation without the need for complex adaptive optics. These findings highlight the potential of photonic lanterns as robust platforms for controlled mode interference, paving the way for their deployment in next-generation optical communication, beam shaping, and beam-forming systems. | [
{
"version": "v1",
"created": "Tue, 10 Jun 2025 09:03:35 GMT"
},
{
"version": "v2",
"created": "Wed, 11 Jun 2025 09:04:20 GMT"
}
] | 2025-06-12T00:00:00 | [
[
"Chandrasekharan",
"Harikumar K",
""
],
[
"Donaldson",
"Ross",
""
]
] |
2506.08622 | Zhu Kaien | Cuizhi Zhou, Kaien Zhu | Physics-Informed Neural Networks for Irregular Domain Mapping and Partial Differential Equations solving | null | null | null | null | physics.comp-ph | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | The solution of partial differential equations (PDES) on irregular domains has long been a subject of significant research interest. In this work, we present an approach utilizing physics-informed neural networks (PINNs) to achieve irregular-to-regular domain mapping. Thus we can use finite difference method and physics-informed convolutional neural networks to solve PDEs on rectangular grids instead of the original irregular boundary.
Structured grids on irregular domains are obtained by inverse mapping. We demonstrate PINN's versatile capability to produce customized structured grids tailored to diverse computational requirements, thereby significantly facilitating PDEs solving. | [
{
"version": "v1",
"created": "Tue, 10 Jun 2025 09:32:13 GMT"
},
{
"version": "v2",
"created": "Wed, 11 Jun 2025 02:09:47 GMT"
}
] | 2025-06-12T00:00:00 | [
[
"Zhou",
"Cuizhi",
""
],
[
"Zhu",
"Kaien",
""
]
] |
2506.08626 | Xingzhu Wang | Xingzhu Wang, Erhan Zhang, Yiqun Chen, Jinghan Xuan, Yucheng Hou, Yitong Xu, Ying Nie, Shuaiqiang Wang, Dawei Yin, Jiaxin Mao | Leveraging LLMs to Evaluate Usefulness of Document | null | null | null | null | cs.IR | http://creativecommons.org/licenses/by/4.0/ | The conventional Cranfield paradigm struggles to effectively capture user satisfaction due to its weak correlation between relevance and satisfaction, alongside the high costs of relevance annotation in building test collections. To tackle these issues, our research explores the potential of leveraging large language models (LLMs) to generate multilevel usefulness labels for evaluation. We introduce a new user-centric evaluation framework that integrates users' search context and behavioral data into LLMs. This framework uses a cascading judgment structure designed for multilevel usefulness assessments, drawing inspiration from ordinal regression techniques. Our study demonstrates that when well-guided with context and behavioral information, LLMs can accurately evaluate usefulness, allowing our approach to surpass third-party labeling methods. Furthermore, we conduct ablation studies to investigate the influence of key components within the framework. We also apply the labels produced by our method to predict user satisfaction, with real-world experiments indicating that these labels substantially improve the performance of satisfaction prediction models. | [
{
"version": "v1",
"created": "Tue, 10 Jun 2025 09:44:03 GMT"
},
{
"version": "v2",
"created": "Wed, 11 Jun 2025 03:47:20 GMT"
}
] | 2025-06-12T00:00:00 | [
[
"Wang",
"Xingzhu",
""
],
[
"Zhang",
"Erhan",
""
],
[
"Chen",
"Yiqun",
""
],
[
"Xuan",
"Jinghan",
""
],
[
"Hou",
"Yucheng",
""
],
[
"Xu",
"Yitong",
""
],
[
"Nie",
"Ying",
""
],
[
"Wang",
"Shuaiqiang",
""
],
[
"Yin",
"Dawei",
""
],
[
"Mao",
"Jiaxin",
""
]
] |
2506.08650 | Peter Gr\"onquist | Peter Gr\"onquist, Stepan Tulyakov, Dengxin Dai | Beyond Calibration: Physically Informed Learning for Raw-to-Raw Mapping | null | null | null | null | cs.CV | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Achieving consistent color reproduction across multiple cameras is essential for seamless image fusion and Image Processing Pipeline (ISP) compatibility in modern devices, but it is a challenging task due to variations in sensors and optics. Existing raw-to-raw conversion methods face limitations such as poor adaptability to changing illumination, high computational costs, or impractical requirements such as simultaneous camera operation and overlapping fields-of-view. We introduce the Neural Physical Model (NPM), a lightweight, physically-informed approach that simulates raw images under specified illumination to estimate transformations between devices. The NPM effectively adapts to varying illumination conditions, can be initialized with physical measurements, and supports training with or without paired data. Experiments on public datasets like NUS and BeyondRGB demonstrate that NPM outperforms recent state-of-the-art methods, providing robust chromatic consistency across different sensors and optical systems. | [
{
"version": "v1",
"created": "Tue, 10 Jun 2025 10:02:00 GMT"
},
{
"version": "v2",
"created": "Wed, 11 Jun 2025 03:36:46 GMT"
}
] | 2025-06-12T00:00:00 | [
[
"Grönquist",
"Peter",
""
],
[
"Tulyakov",
"Stepan",
""
],
[
"Dai",
"Dengxin",
""
]
] |
2506.08657 | Aniket Patra | Aniket Patra, Emil A. Yuzbashyan, Boris L. Altshuler, and Sergej Flach | Trotter transition in BCS pairing dynamics | 16 pages, 13 figures | null | null | null | quant-ph cond-mat.supr-con nlin.CD | http://creativecommons.org/licenses/by/4.0/ | We Trotterize the mean-field dynamics of the integrable BCS model using symplectic integrators. The resulting chaotic dynamics is characterized by its Lyapunov spectrum and rescaled Kolmogorov-Sinai entropy. The chaos quantifiers depend on the Trotterization time step $\tau$. We observe a Trotter transition at a finite step value $\tau_c \approx \sqrt{N}$. While the dynamics is weakly chaotic for time steps $\tau \ll \tau_c$, the regime of large Trotterization steps is characterized by short temporal correlations. We derive two different scaling laws for the two different regimes by numerically fitting the maximum Lyapunov exponent data. The scaling law of the large $\tau$ limit agrees well with the one derived from the kicked top map. Beyond its relevance to current quantum computers, our work opens new directions -- such as probing observables like the Loschmidt echo, which lie beyond standard mean-field description -- across the Trotter transition we uncover. | [
{
"version": "v1",
"created": "Tue, 10 Jun 2025 10:19:10 GMT"
}
] | 2025-06-12T00:00:00 | [
[
"Patra",
"Aniket",
""
],
[
"Yuzbashyan",
"Emil A.",
""
],
[
"Altshuler",
"Boris L.",
""
],
[
"Flach",
"Sergej",
""
]
] |
2506.08672 | Yang Liu | Yang Liu, Jiaqi Li, Zilong Zheng | RuleReasoner: Reinforced Rule-based Reasoning via Domain-aware Dynamic Sampling | 22 pages, 10 figures, 8 tables | null | null | null | cs.CL cs.AI cs.LG | http://creativecommons.org/licenses/by-nc-nd/4.0/ | Rule-based reasoning has been acknowledged as one of the fundamental problems in reasoning, while deviations in rule formats, types, and complexity in real-world applications pose severe challenges. Recent studies have shown that large reasoning models (LRMs) have remarkable reasoning capabilities, and their performance is substantially enhanced by reinforcement learning (RL). However, it remains an open question whether small reasoning models (SRMs) can learn rule-based reasoning effectively with robust generalization across diverse tasks and domains. To address this, we introduce Reinforced Rule-based Reasoning, a.k.a. RuleReasoner, a simple yet effective method to conduct rule-based reasoning via a wide collection of curated tasks and a novel domain-aware dynamic sampling approach. Specifically, RuleReasoner resamples each training batch by updating the sampling weights of different domains based on historical rewards. This facilitates domain augmentation and flexible online learning schedules for RL, obviating the need for pre-hoc human-engineered mix-training recipes used in existing methods. Empirical evaluations on in-distribution (ID) and out-of-distribution (OOD) benchmarks reveal that RuleReasoner outperforms frontier LRMs by a significant margin ($\Delta$4.1% average points on eight ID tasks and $\Delta$10.4% average points on three OOD tasks over OpenAI-o1). Notably, our approach also exhibits higher computational efficiency compared to prior dynamic sampling methods for RL. | [
{
"version": "v1",
"created": "Tue, 10 Jun 2025 10:31:21 GMT"
}
] | 2025-06-12T00:00:00 | [
[
"Liu",
"Yang",
""
],
[
"Li",
"Jiaqi",
""
],
[
"Zheng",
"Zilong",
""
]
] |
2506.08674 | Karolis Daugevi\v{c}ius | Karolis Daugevi\v{c}ius, Eimantas Kri\v{s}\v{c}i\=unas, Erikas Cic\.enas, Rima Stonkut\.e, and Vladas Vansevi\v{c}ius | Deriving physical parameters of unresolved star clusters. VIII. Limits of aperture photometry for star cluster studies | 14 pages, 15 figures, published in A&A | Astronomy & Astrophysics, 688, A131 (2024) | 10.1051/0004-6361/202449680 | null | astro-ph.GA | http://creativecommons.org/licenses/by/4.0/ | Recently, it has been noticed that the discrepancies in the integrated colour indices (CIs) between star clusters and models are mostly due to the projection of bright stars in the apertures. In order to reduce this problem, the method of adaptive aperture photometry has been proposed. This method has been applied to star clusters from the M 31 Panchromatic Hubble Andromeda Treasury (PHAT) survey, and studies show that the adaptive aperture photometry performs better than the conventional approach. The aim of this study is to determine the best achievable limits on the accuracy and applicability of the aperture photometry method for studying star clusters in the local Universe. We computed a large network of artificial 3D star clusters spanning the parameter space of the M 31 clusters. We then simulated images of these clusters by projecting each onto a 2D plane from 100 directions. Star cluster images were generated in six passbands to match the PHAT survey. To investigate the limiting accuracy of aperture photometry and the limits of its applicability to star cluster studies, we measured the simulated images and performed parameter determination tests. We demonstrate that star clusters with and without post-main-sequence stars have significant photometric differences. We show that in order to obtain reliable physical parameters of star clusters, the CIs must be measured using an aperture with a radius larger than the cluster's half-light radius. Furthermore, we demonstrate that the parameter determination of young clusters (~10 Myr) is problematic regardless of the aperture size used. Therefore, it is advisable to determine the parameters of these clusters using colour-magnitude diagram fitting methods, when possible. We also show that the randomness of the viewing angle can lead to a CI uncertainty of up to 0.1 mag, depending on cluster parameters and aperture size. | [
{
"version": "v1",
"created": "Tue, 10 Jun 2025 10:33:47 GMT"
}
] | 2025-06-12T00:00:00 | [
[
"Daugevičius",
"Karolis",
""
],
[
"Kriščiūnas",
"Eimantas",
""
],
[
"Cicėnas",
"Erikas",
""
],
[
"Stonkutė",
"Rima",
""
],
[
"Vansevičius",
"Vladas",
""
]
] |
2506.08681 | Duy Minh Ho Nguyen | Phuc Minh Nguyen, Ngoc-Hieu Nguyen, Duy H. M. Nguyen, Anji Liu, An Mai, Binh T. Nguyen, Daniel Sonntag, Khoa D. Doan | Mitigating Reward Over-optimization in Direct Alignment Algorithms with Importance Sampling | First version | null | null | null | cs.LG | http://creativecommons.org/licenses/by/4.0/ | Direct Alignment Algorithms (DAAs) such as Direct Preference Optimization (DPO) have emerged as alternatives to the standard Reinforcement Learning from Human Feedback (RLHF) for aligning large language models (LLMs) with human values. However, these methods are more susceptible to over-optimization, in which the model drifts away from the reference policy, leading to degraded performance as training progresses. This paper proposes a novel importance-sampling approach to mitigate the over-optimization problem of offline DAAs. This approach, called (IS-DAAs), multiplies the DAA objective with an importance ratio that accounts for the reference policy distribution. IS-DAAs additionally avoid the high variance issue associated with importance sampling by clipping the importance ratio to a maximum value. Our extensive experiments demonstrate that IS-DAAs can effectively mitigate over-optimization, especially under low regularization strength, and achieve better performance than other methods designed to address this problem. Our implementations are provided publicly at this link. | [
{
"version": "v1",
"created": "Tue, 10 Jun 2025 10:45:26 GMT"
},
{
"version": "v2",
"created": "Wed, 11 Jun 2025 06:36:33 GMT"
}
] | 2025-06-12T00:00:00 | [
[
"Nguyen",
"Phuc Minh",
""
],
[
"Nguyen",
"Ngoc-Hieu",
""
],
[
"Nguyen",
"Duy H. M.",
""
],
[
"Liu",
"Anji",
""
],
[
"Mai",
"An",
""
],
[
"Nguyen",
"Binh T.",
""
],
[
"Sonntag",
"Daniel",
""
],
[
"Doan",
"Khoa D.",
""
]
] |
2506.08700 | Ruiran Su | Ruiran Su, Jiasheng Si, Zhijiang Guo, Janet B. Pierrehumbert | ClimateViz: A Benchmark for Statistical Reasoning and Fact Verification on Scientific Charts | null | null | null | null | cs.CL cs.CV | http://creativecommons.org/licenses/by/4.0/ | Scientific fact-checking has mostly focused on text and tables, overlooking scientific charts, which are key for presenting quantitative evidence and statistical reasoning. We introduce ClimateViz, the first large-scale benchmark for scientific fact-checking using expert-curated scientific charts. ClimateViz contains 49,862 claims linked to 2,896 visualizations, each labeled as support, refute, or not enough information. To improve interpretability, each example includes structured knowledge graph explanations covering trends, comparisons, and causal relations. We evaluate state-of-the-art multimodal language models, including both proprietary and open-source systems, in zero-shot and few-shot settings. Results show that current models struggle with chart-based reasoning: even the best systems, such as Gemini 2.5 and InternVL 2.5, reach only 76.2 to 77.8 percent accuracy in label-only settings, far below human performance (89.3 and 92.7 percent). Explanation-augmented outputs improve performance in some models. We released our dataset and code alongside the paper. | [
{
"version": "v1",
"created": "Tue, 10 Jun 2025 11:27:00 GMT"
},
{
"version": "v2",
"created": "Wed, 11 Jun 2025 09:49:01 GMT"
}
] | 2025-06-12T00:00:00 | [
[
"Su",
"Ruiran",
""
],
[
"Si",
"Jiasheng",
""
],
[
"Guo",
"Zhijiang",
""
],
[
"Pierrehumbert",
"Janet B.",
""
]
] |
2506.08703 | Victor Rueskov Christiansen | Victor Rueskov Christiansen and Klaus M{\o}lmer | Interactions in Quantum Networks with Pulse Propagation Delays | 8 pages, 4 figures | null | null | null | quant-ph | http://creativecommons.org/licenses/by/4.0/ | The finite speed of light of pulses implies propagation delays between the generation of light, the interaction with different elements along its path and the final detection at the output of networks and interferometers. We show that it is possible to take these delays into account with a theoretical method which evades quantization of the full continuum of field modes. We illustrate the use of this method by analyzing Ramsey excitation of an atom by a split and delayed quantum pulse. | [
{
"version": "v1",
"created": "Tue, 10 Jun 2025 11:30:59 GMT"
},
{
"version": "v2",
"created": "Wed, 11 Jun 2025 11:09:09 GMT"
}
] | 2025-06-12T00:00:00 | [
[
"Christiansen",
"Victor Rueskov",
""
],
[
"Mølmer",
"Klaus",
""
]
] |
2506.08720 | Fr\'ed\'eric Zheng | Fr\'ed\'eric Zheng, Yassir Jedra, Alexandre Prouti\`ere | Minimal Order Recovery through Rank-adaptive Identification | null | null | null | null | eess.SY cs.SY | http://creativecommons.org/licenses/by/4.0/ | This paper addresses the problem of identifying linear systems from noisy input-output trajectories. We introduce Thresholded Ho-Kalman, an algorithm that leverages a rank-adaptive procedure to estimate a Hankel-like matrix associated with the system. This approach optimally balances the trade-off between accurately inferring key singular values and minimizing approximation errors for the rest. We establish finite-sample Frobenius norm error bounds for the estimated Hankel matrix. Our algorithm further recovers both the system order and its Markov parameters, and we provide upper bounds for the sample complexity required to identify the system order and finite-time error bounds for estimating the Markov parameters. Interestingly, these bounds match those achieved by state-of-the-art algorithms that assume prior knowledge of the system order. | [
{
"version": "v1",
"created": "Tue, 10 Jun 2025 12:12:49 GMT"
},
{
"version": "v2",
"created": "Wed, 11 Jun 2025 07:29:27 GMT"
}
] | 2025-06-12T00:00:00 | [
[
"Zheng",
"Frédéric",
""
],
[
"Jedra",
"Yassir",
""
],
[
"Proutière",
"Alexandre",
""
]
] |
2506.08729 | Dieuwertje Alblas | Dieuwertje Alblas, Patryk Rygiel, Julian Suk, Kaj O. Kappe, Marieke Hofman, Christoph Brune, Kak Khee Yeung, Jelmer M. Wolterink | Geometric deep learning for local growth prediction on abdominal aortic aneurysm surfaces | null | null | null | null | cs.CV cs.AI | http://creativecommons.org/licenses/by/4.0/ | Abdominal aortic aneurysms (AAAs) are progressive focal dilatations of the abdominal aorta. AAAs may rupture, with a survival rate of only 20\%. Current clinical guidelines recommend elective surgical repair when the maximum AAA diameter exceeds 55 mm in men or 50 mm in women. Patients that do not meet these criteria are periodically monitored, with surveillance intervals based on the maximum AAA diameter. However, this diameter does not take into account the complex relation between the 3D AAA shape and its growth, making standardized intervals potentially unfit. Personalized AAA growth predictions could improve monitoring strategies. We propose to use an SE(3)-symmetric transformer model to predict AAA growth directly on the vascular model surface enriched with local, multi-physical features. In contrast to other works which have parameterized the AAA shape, this representation preserves the vascular surface's anatomical structure and geometric fidelity. We train our model using a longitudinal dataset of 113 computed tomography angiography (CTA) scans of 24 AAA patients at irregularly sampled intervals. After training, our model predicts AAA growth to the next scan moment with a median diameter error of 1.18 mm. We further demonstrate our model's utility to identify whether a patient will become eligible for elective repair within two years (acc = 0.93). Finally, we evaluate our model's generalization on an external validation set consisting of 25 CTAs from 7 AAA patients from a different hospital. Our results show that local directional AAA growth prediction from the vascular surface is feasible and may contribute to personalized surveillance strategies. | [
{
"version": "v1",
"created": "Tue, 10 Jun 2025 12:27:12 GMT"
},
{
"version": "v2",
"created": "Wed, 11 Jun 2025 13:26:10 GMT"
}
] | 2025-06-12T00:00:00 | [
[
"Alblas",
"Dieuwertje",
""
],
[
"Rygiel",
"Patryk",
""
],
[
"Suk",
"Julian",
""
],
[
"Kappe",
"Kaj O.",
""
],
[
"Hofman",
"Marieke",
""
],
[
"Brune",
"Christoph",
""
],
[
"Yeung",
"Kak Khee",
""
],
[
"Wolterink",
"Jelmer M.",
""
]
] |
2506.08738 | Dror Markus | Dror Kris Markus, Fabrizio Gilardi, Daria Stetsenko | Societal AI Research Has Become Less Interdisciplinary | null | null | null | null | cs.CL cs.AI cs.CY | http://creativecommons.org/licenses/by/4.0/ | As artificial intelligence (AI) systems become deeply embedded in everyday life, calls to align AI development with ethical and societal values have intensified. Interdisciplinary collaboration is often championed as a key pathway for fostering such engagement. Yet it remains unclear whether interdisciplinary research teams are actually leading this shift in practice. This study analyzes over 100,000 AI-related papers published on ArXiv between 2014 and 2024 to examine how ethical values and societal concerns are integrated into technical AI research. We develop a classifier to identify societal content and measure the extent to which research papers express these considerations. We find a striking shift: while interdisciplinary teams remain more likely to produce societally-oriented research, computer science-only teams now account for a growing share of the field's overall societal output. These teams are increasingly integrating societal concerns into their papers and tackling a wide range of domains - from fairness and safety to healthcare and misinformation. These findings challenge common assumptions about the drivers of societal AI and raise important questions. First, what are the implications for emerging understandings of AI safety and governance if most societally-oriented research is being undertaken by exclusively technical teams? Second, for scholars in the social sciences and humanities: in a technical field increasingly responsive to societal demands, what distinctive perspectives can we still offer to help shape the future of AI? | [
{
"version": "v1",
"created": "Tue, 10 Jun 2025 12:34:53 GMT"
},
{
"version": "v2",
"created": "Wed, 11 Jun 2025 09:42:30 GMT"
}
] | 2025-06-12T00:00:00 | [
[
"Markus",
"Dror Kris",
""
],
[
"Gilardi",
"Fabrizio",
""
],
[
"Stetsenko",
"Daria",
""
]
] |
2506.08748 | Ivo Mihov | Ivo S. Mihov and Nikolay V. Vitanov | Observation of Power Superbroadening of Spectral Line Profiles on IBM Quantum | 7 pages, 5 figures (1 figure and 2 x 2 subfigures) | null | null | null | quant-ph | http://creativecommons.org/licenses/by/4.0/ | Power broadening refers to the widening of the spectral line profile in a two-state quantum transition as the strength of the driving field increases. This phenomenon commonly arises in continuous-wave driving when the radiation field's intensity exceeds the transition's saturation intensity and it has been extensively studied in spectroscopy. For pulsed-field excitation, the spectral response of the quantum system may differ significantly: while a rectangular-shaped pulse leads to a linear power broadening, pulses with smooth shapes show significantly reduced power broadening, for instance, logarithmic for the Gaussian shape and none for the hyperbolic-secant shape. Recently [Phys. Rev. Lett. 132, 020802 (2024)], in a dramatic paradigm shift, we have demonstrated experimentally that for Lorentzian-shaped pulses, the opposite effect - power narrowing - takes place: the width of the spectral profile decreases when the driving pulse amplitude increases, with a narrowing factor of as much as 10 observed.
While in high-resolution spectroscopy the push is for eliminating or even inverting the power broadening, there are applications where it is used to an advantage for it facilitates off-resonance excitation. Here, we present a number of shaped pulses that exhibit power broadening much greater than that of the rectangular pulse of the same pulse area. They are grouped in two families of pulse shapes. In particular, in regard to the width of the second Rabi oscillation maximum, the quadratic pulse family shows an increase by a factor of 3.3 whereas the even-exponent pulse family exhibits an increase by a factor of more than 3.5. | [
{
"version": "v1",
"created": "Tue, 10 Jun 2025 12:43:25 GMT"
},
{
"version": "v2",
"created": "Wed, 11 Jun 2025 08:11:32 GMT"
}
] | 2025-06-12T00:00:00 | [
[
"Mihov",
"Ivo S.",
""
],
[
"Vitanov",
"Nikolay V.",
""
]
] |
2506.08765 | Hamidreza Imani | Hamid R. Imani, K. Babaei Velni, M. Reza Mohammadi Mozaffar | Krylov Complexity in Lifshitz-type Dirac Field Theories | 26 pages, 18 figures, v2: references added | null | null | null | hep-th quant-ph | http://creativecommons.org/licenses/by-nc-nd/4.0/ | We study Krylov complexity in Lifshitz-type Dirac field theories with a generic dynamical critical exponent z. By computing the Lanczos coefficients for massless and massive cases, we analyze the growth and saturation behavior of Krylov complexity in different regimes. We incorporate a hard UV cutoff and investigate the effects of lattice discretization, revealing fundamental differences between continuum and lattice models. In the presence of a UV cutoff, Krylov complexity exhibits an initial exponential growth followed by a linear regime, with saturation values dictated by the cutoff scale. For the lattice model, we find a fundamental departure from the continuum case: due to the finite Krylov basis, Krylov complexity saturates rather than growing indefinitely. Our findings suggest that Lifshitz scaling influences operator growth and information spreading in quantum systems. | [
{
"version": "v1",
"created": "Tue, 10 Jun 2025 13:05:08 GMT"
},
{
"version": "v2",
"created": "Wed, 11 Jun 2025 13:21:01 GMT"
}
] | 2025-06-12T00:00:00 | [
[
"Imani",
"Hamid R.",
""
],
[
"Velni",
"K. Babaei",
""
],
[
"Mozaffar",
"M. Reza Mohammadi",
""
]
] |
2506.08768 | Aisha Alansari | Ahmed Hasanaath, Aisha Alansari, Ahmed Ashraf, Chafik Salmane, Hamzah Luqman, Saad Ezzini | AraReasoner: Evaluating Reasoning-Based LLMs for Arabic NLP | null | null | null | null | cs.CL | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Large language models (LLMs) have shown remarkable progress in reasoning abilities and general natural language processing (NLP) tasks, yet their performance on Arabic data, characterized by rich morphology, diverse dialects, and complex script, remains underexplored. This paper presents a comprehensive benchmarking study of multiple reasoning-focused LLMs, with a special emphasis on the newly introduced DeepSeek models, across a suite of fifteen Arabic NLP tasks. We experiment with various strategies, including zero-shot, few-shot, and fine-tuning. This allows us to systematically evaluate performance on datasets covering a range of applications to examine their capacity for linguistic reasoning under different levels of complexity. Our experiments reveal several key findings. First, carefully selecting just three in-context examples delivers an average uplift of over 13 F1 points on classification tasks-boosting sentiment analysis from 35.3% to 87.5% and paraphrase detection from 56.1% to 87.0%. Second, reasoning-focused DeepSeek architectures outperform a strong GPT o4-mini baseline by an average of 12 F1 points on complex inference tasks in the zero-shot setting. Third, LoRA-based fine-tuning yields up to an additional 8 points in F1 and BLEU compared to equivalent increases in model scale. The code is available at https://anonymous.4open.science/r/AraReasoner41299 | [
{
"version": "v1",
"created": "Tue, 10 Jun 2025 13:10:31 GMT"
},
{
"version": "v2",
"created": "Wed, 11 Jun 2025 09:00:02 GMT"
}
] | 2025-06-12T00:00:00 | [
[
"Hasanaath",
"Ahmed",
""
],
[
"Alansari",
"Aisha",
""
],
[
"Ashraf",
"Ahmed",
""
],
[
"Salmane",
"Chafik",
""
],
[
"Luqman",
"Hamzah",
""
],
[
"Ezzini",
"Saad",
""
]
] |
2506.08769 | Ryan Arbon | Ryan Arbon and Jacob Bedrossian | Stability of Stochastically Driven Couette Flow in 2D with Navier Boundary Conditions at high Reynolds number via Averaging Principle | null | null | null | null | math.AP math.PR | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | We characterize the behavior of stochastic Navier-Stokes on $\mathbb{T} \times [-1,1]$ with Navier boundary conditions at high Reynolds number when initialized near Couette flow subject to small additive stochastic forcing. We take additive noise of strength $\nu^{1/2+} \Phi dV_t + \nu^{2/3+\alpha} \Psi dW_t$, where $\Phi dV_t$ has spatial correlation in $H_0^3$ and acts only on $x$-independent modes of the vorticity, while $\Psi dW_t$ has spatial correlation in a lower order, anisotropic, Sobolev space $\mathcal{H}$ and acts on $x$-dependent-modes. We take the initial $x$-independent modes in the perturbation to be small in $H_0^3$ in a $\nu$-independent sense, while the non-zero $x$-modes are taken to be $O(\nu^{1/2 + \alpha})$ in $\mathcal{H}$. Letting $\omega$ solve the resulting perturbation equation, we split $\omega$ into the zero $x$-modes $\omega_0$ and the non-zero $x$-modes $\omega_{\neq}$. We demonstrate an averaging principle holds wherein $\omega_{\neq}$ is the fast variable and $\omega_0$ is the slow variable, deriving a closed nonlinear evolution equation on $\omega_0$ that holds over long time-scales (while the fast $\omega_{\neq}$ modes solve a `pseudo-linearized' equation to leading order with dynamics dominated by inviscid damping and enhanced dissipation). This work can also be considered the stochastic analogue of the stability threshold problem for shear flows. | [
{
"version": "v1",
"created": "Tue, 10 Jun 2025 13:11:06 GMT"
}
] | 2025-06-12T00:00:00 | [
[
"Arbon",
"Ryan",
""
],
[
"Bedrossian",
"Jacob",
""
]
] |
2506.08772 | Kaiyu Li | Jiayi Song, Kaiyu Li, Xiangyong Cao, Deyu Meng | RS-MTDF: Multi-Teacher Distillation and Fusion for Remote Sensing Semi-Supervised Semantic Segmentation | null | null | null | null | cs.CV | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Semantic segmentation in remote sensing images is crucial for various applications, yet its performance is heavily reliant on large-scale, high-quality pixel-wise annotations, which are notoriously expensive and time-consuming to acquire. Semi-supervised semantic segmentation (SSS) offers a promising alternative to mitigate this data dependency. However, existing SSS methods often struggle with the inherent distribution mismatch between limited labeled data and abundant unlabeled data, leading to suboptimal generalization. To alleviate this issue, we attempt to introduce the Vision Foundation Models (VFMs) pre-trained on vast and diverse datasets into the SSS task since VFMs possess robust generalization capabilities that can effectively bridge this distribution gap and provide strong semantic priors for SSS. Inspired by this, we introduce RS-MTDF (Multi-Teacher Distillation and Fusion), a novel framework that leverages the powerful semantic knowledge embedded in VFMs to guide semi-supervised learning in remote sensing. Specifically, RS-MTDF employs multiple frozen VFMs (e.g., DINOv2 and CLIP) as expert teachers, utilizing feature-level distillation to align student features with their robust representations. To further enhance discriminative power, the distilled knowledge is seamlessly fused into the student decoder. Extensive experiments on three challenging remote sensing datasets demonstrate that RS-MTDF consistently achieves state-of-the-art performance. Notably, our method outperforms existing approaches across various label ratios on LoveDA and secures the highest IoU in the majority of semantic categories. These results underscore the efficacy of multi-teacher VFM guidance in significantly enhancing both generalization and semantic understanding for remote sensing segmentation. Ablation studies further validate the contribution of each proposed module. | [
{
"version": "v1",
"created": "Tue, 10 Jun 2025 13:15:15 GMT"
},
{
"version": "v2",
"created": "Wed, 11 Jun 2025 16:46:53 GMT"
}
] | 2025-06-12T00:00:00 | [
[
"Song",
"Jiayi",
""
],
[
"Li",
"Kaiyu",
""
],
[
"Cao",
"Xiangyong",
""
],
[
"Meng",
"Deyu",
""
]
] |
2506.08777 | Keyi Liu | Keyi Liu and Weidong Yang and Ben Fei and Ying He | Gaussian2Scene: 3D Scene Representation Learning via Self-supervised Learning with 3D Gaussian Splatting | null | null | null | null | cs.CV | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Self-supervised learning (SSL) for point cloud pre-training has become a cornerstone for many 3D vision tasks, enabling effective learning from large-scale unannotated data. At the scene level, existing SSL methods often incorporate volume rendering into the pre-training framework, using RGB-D images as reconstruction signals to facilitate cross-modal learning. This strategy promotes alignment between 2D and 3D modalities and enables the model to benefit from rich visual cues in the RGB-D inputs. However, these approaches are limited by their reliance on implicit scene representations and high memory demands. Furthermore, since their reconstruction objectives are applied only in 2D space, they often fail to capture underlying 3D geometric structures. To address these challenges, we propose Gaussian2Scene, a novel scene-level SSL framework that leverages the efficiency and explicit nature of 3D Gaussian Splatting (3DGS) for pre-training. The use of 3DGS not only alleviates the computational burden associated with volume rendering but also supports direct 3D scene reconstruction, thereby enhancing the geometric understanding of the backbone network. Our approach follows a progressive two-stage training strategy. In the first stage, a dual-branch masked autoencoder learns both 2D and 3D scene representations. In the second stage, we initialize training with reconstructed point clouds and further supervise learning using the geometric locations of Gaussian primitives and rendered RGB images. This process reinforces both geometric and cross-modal learning. We demonstrate the effectiveness of Gaussian2Scene across several downstream 3D object detection tasks, showing consistent improvements over existing pre-training methods. | [
{
"version": "v1",
"created": "Tue, 10 Jun 2025 13:19:21 GMT"
},
{
"version": "v2",
"created": "Wed, 11 Jun 2025 08:42:02 GMT"
}
] | 2025-06-12T00:00:00 | [
[
"Liu",
"Keyi",
""
],
[
"Yang",
"Weidong",
""
],
[
"Fei",
"Ben",
""
],
[
"He",
"Ying",
""
]
] |
2506.08779 | Dor Shohat | Dor Shohat, Paul Baconnier, Itamar Procaccia, Martin van Hecke, and Yoav Lahini | Aging of amorphous materials under cyclic strain | 13 pages, 7 figures | null | null | null | cond-mat.soft cond-mat.dis-nn | http://creativecommons.org/licenses/by/4.0/ | Amorphous materials driven away from equilibrium display a diverse repertoire of complex, history-dependent behaviors. One striking feature is a failure to return to equilibrium after an abrupt change in otherwise static external conditions. Instead, amorphous materials often exhibit physical aging: an ever-slowing, nonexponential relaxation that can span a huge range of timescales. Here we examine the aging behavior of three different amorphous materials subjected to slow periodic driving. The results reveal a generic aging phenomenon characterized by a logarithmic decay of dissipation per cycle. This observation is evaluated against several mesoscopic models of amorphous matter that successfully capture aging under static conditions: (i) a collection of noninteracting relaxation processes (ii) a noisy hysteron model with random pairwise interactions, and (iii) a structural model consisting of a random network of bi-stable elastic bonds. We find that only the latter model reproduces all experimental findings and relate its success to its persistent, slow exploration of a complex energy landscape with clear signatures of replica symmetry breaking. Thus, cyclic driving emerges as a simple yet powerful protocol to characterize amorphous materials, probe their complex energy landscapes, and distinguish between different models. | [
{
"version": "v1",
"created": "Tue, 10 Jun 2025 13:23:56 GMT"
}
] | 2025-06-12T00:00:00 | [
[
"Shohat",
"Dor",
""
],
[
"Baconnier",
"Paul",
""
],
[
"Procaccia",
"Itamar",
""
],
[
"van Hecke",
"Martin",
""
],
[
"Lahini",
"Yoav",
""
]
] |
2506.08803 | Rolf Schneider | Rolf Schneider | Anisotropic area measures of convex bodies | 12 pages | null | null | null | math.MG | http://creativecommons.org/licenses/by/4.0/ | Motivated by the relative differential geometry, where the Euclidean normal vector of hypersurfaces is generalized by a relative normalization, we introduce anisotropic area measures of convex bodies, constructed with respect to a gauge body. Together with the anisotropic curvature measures, they are special cases of the newly introduced anisotropic support measures. We show that a convex body in ${\mathbb R}^n$, for which the anisotropic area measure of some order $k\in\{0,\dots,n-2\}$ is proportional to the area measure of order $n-1$, must be a $k$-tangential body of the gauge body. | [
{
"version": "v1",
"created": "Tue, 10 Jun 2025 13:50:11 GMT"
},
{
"version": "v2",
"created": "Wed, 11 Jun 2025 12:08:19 GMT"
}
] | 2025-06-12T00:00:00 | [
[
"Schneider",
"Rolf",
""
]
] |
2506.08804 | Michael Heines | K.A. Beyer, T.E. Cocolios, C. Costache, M. Deseyn, P. Demol, A. Doinaki, O. Eizenberg, M. Gorshteyn, M. Heines, A. Herz\'a\v{n}, P. Indelicato, K. Kirch, A. Knecht, R. Lica, V. Matousek, E.A. Maugeri, B. Ohayon, N.S. Oreshkina, W.W.M.M. Phyo, R. Pohl, S. Rathi, W. Ryssens, A. Turturica, K. von Schoeler, I.A. Valuev, S.M. Vogiatzi, F. Wauters, A. Zendour | Modern approach to muonic x-ray spectroscopy demonstrated through the measurement of stable Cl radii | null | null | null | null | nucl-ex nucl-th physics.atom-ph physics.data-an | http://creativecommons.org/licenses/by/4.0/ | Recent advances in muonic x-ray experiments have reinvigorated efforts in measurements of absolute nuclear charge radii. Here, a modern approach is presented, and demonstrated through determination of the charge radii of the two stable chlorine nuclides $^{35}$Cl and $^{37}$Cl. Knowledge of these radii has implications for fundamental studies in nuclear and atomic physics. For this purpose, a state-of-the-art experiment was performed at the $\pi$E1 beamline in the Paul Scherrer Institute (Switzerland), using a large-scale HPGe detector array in order to extract precise energies of the muonic $^{35}$Cl and $^{37}$Cl $np1s$ transitions. The nuclear charge radius extraction relies on modern calculations for QED effects and nuclear polarization with rigorous uncertainty quantification, including effects that were not accounted for in older studies. Additionally, we established a new method for applying the nuclear shape correction directly from energy density functionals, which are amenable to isotopes for which no high-quality electron scattering experiments are available. The resulting charge radii are $3.3335(23) fm$ for $^{35}$Cl and $3.3445(23) fm$ for $^{37}$Cl, thus improving the uncertainty of the available electron scattering values by a factor of seven. The correlation of several observables was evaluated between the different isotopes in order to produce a more precise value of the differential mean square charge radius $\delta \langle r^2 \rangle^{37, 35}=+0.0771(66) fm^{2}$. In this case, improvement of the uncertainty by more than one order of magnitude was achieved compared to the literature value. This precision is sufficient to use this differential as input for isotope shift factor determination. | [
{
"version": "v1",
"created": "Tue, 10 Jun 2025 13:50:18 GMT"
},
{
"version": "v2",
"created": "Wed, 11 Jun 2025 06:55:33 GMT"
}
] | 2025-06-12T00:00:00 | [
[
"Beyer",
"K. A.",
""
],
[
"Cocolios",
"T. E.",
""
],
[
"Costache",
"C.",
""
],
[
"Deseyn",
"M.",
""
],
[
"Demol",
"P.",
""
],
[
"Doinaki",
"A.",
""
],
[
"Eizenberg",
"O.",
""
],
[
"Gorshteyn",
"M.",
""
],
[
"Heines",
"M.",
""
],
[
"Herzáň",
"A.",
""
],
[
"Indelicato",
"P.",
""
],
[
"Kirch",
"K.",
""
],
[
"Knecht",
"A.",
""
],
[
"Lica",
"R.",
""
],
[
"Matousek",
"V.",
""
],
[
"Maugeri",
"E. A.",
""
],
[
"Ohayon",
"B.",
""
],
[
"Oreshkina",
"N. S.",
""
],
[
"Phyo",
"W. W. M. M.",
""
],
[
"Pohl",
"R.",
""
],
[
"Rathi",
"S.",
""
],
[
"Ryssens",
"W.",
""
],
[
"Turturica",
"A.",
""
],
[
"von Schoeler",
"K.",
""
],
[
"Valuev",
"I. A.",
""
],
[
"Vogiatzi",
"S. M.",
""
],
[
"Wauters",
"F.",
""
],
[
"Zendour",
"A.",
""
]
] |
2506.08812 | Rohit Mehra | Priyavanshi Pathania, Rohit Mehra, Vibhu Saujanya Sharma, Vikrant Kaulgud, Sanjay Podder, Adam P. Burden | Towards a Knowledge Base of Common Sustainability Weaknesses in Green Software Development | 3 pages. To be published in the proceedings of 38th IEEE/ACM International Conference on Automated Software Engineering (ASE 2023), Kirchberg, Luxembourg | null | 10.1109/ASE56229.2023.00204 | null | cs.SE cs.CY | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | With the climate crisis looming, engineering sustainable software systems become crucial to optimize resource utilization, minimize environmental impact, and foster a greener, more resilient digital ecosystem. For developers, getting access to automated tools that analyze code and suggest sustainabilityrelated optimizations becomes extremely important from a learning and implementation perspective. However, there is currently a dearth of such tools due to the lack of standardized knowledge, which serves as the foundation of these tools. In this paper, we motivate the need for the development of a standard knowledge base of commonly occurring sustainability weaknesses in code, and propose an initial way of doing that. Furthermore, through preliminary experiments, we demonstrate why existing knowledge regarding software weaknesses cannot be re-tagged "as is" to sustainability without significant due diligence, thereby urging further explorations in this ecologically significant domain. | [
{
"version": "v1",
"created": "Tue, 10 Jun 2025 14:03:58 GMT"
}
] | 2025-06-12T00:00:00 | [
[
"Pathania",
"Priyavanshi",
""
],
[
"Mehra",
"Rohit",
""
],
[
"Sharma",
"Vibhu Saujanya",
""
],
[
"Kaulgud",
"Vikrant",
""
],
[
"Podder",
"Sanjay",
""
],
[
"Burden",
"Adam P.",
""
]
] |
2506.08837 | Florian Tram\`er | Luca Beurer-Kellner, Beat Buesser Ana-Maria Cre\c{t}u, Edoardo Debenedetti, Daniel Dobos, Daniel Fabian, Marc Fischer, David Froelicher, Kathrin Grosse, Daniel Naeff, Ezinwanne Ozoani, Andrew Paverd, Florian Tram\`er, V\'aclav Volhejn | Design Patterns for Securing LLM Agents against Prompt Injections | null | null | null | null | cs.LG cs.CR | http://creativecommons.org/licenses/by/4.0/ | As AI agents powered by Large Language Models (LLMs) become increasingly versatile and capable of addressing a broad spectrum of tasks, ensuring their security has become a critical challenge. Among the most pressing threats are prompt injection attacks, which exploit the agent's resilience on natural language inputs -- an especially dangerous threat when agents are granted tool access or handle sensitive information. In this work, we propose a set of principled design patterns for building AI agents with provable resistance to prompt injection. We systematically analyze these patterns, discuss their trade-offs in terms of utility and security, and illustrate their real-world applicability through a series of case studies. | [
{
"version": "v1",
"created": "Tue, 10 Jun 2025 14:23:55 GMT"
},
{
"version": "v2",
"created": "Wed, 11 Jun 2025 09:20:58 GMT"
}
] | 2025-06-12T00:00:00 | [
[
"Beurer-Kellner",
"Luca",
""
],
[
"Creţu",
"Beat Buesser Ana-Maria",
""
],
[
"Debenedetti",
"Edoardo",
""
],
[
"Dobos",
"Daniel",
""
],
[
"Fabian",
"Daniel",
""
],
[
"Fischer",
"Marc",
""
],
[
"Froelicher",
"David",
""
],
[
"Grosse",
"Kathrin",
""
],
[
"Naeff",
"Daniel",
""
],
[
"Ozoani",
"Ezinwanne",
""
],
[
"Paverd",
"Andrew",
""
],
[
"Tramèr",
"Florian",
""
],
[
"Volhejn",
"Václav",
""
]
] |
2506.08847 | Jianfeng Lin | Jianfeng Lin, Zhongzi Wang | $\pi_1$-injective bounding and application to 3- and 4-manifolds | null | null | null | null | math.GT | http://creativecommons.org/licenses/by-nc-nd/4.0/ | Suppose a closed oriented $n$-manifold $M$ bounds an oriented $(n+1)$-manifold. It is known that $M$ $\pi_1$-injectively bounds an oriented $(n+1)$-manifold $W$. We prove that $\pi_1(W)$ can be residually finite if $\pi_1(M)$ is, and $\pi_1(W)$ can be finite if $\pi_1(M)$ is. In particular, each closed 3-manifold $M$ $\pi_1$-injectively bounds a 4-manifold with residually finite $\pi_1$, and bounds a 4-manifold with finite $\pi_1$ if $\pi_1(M)$ is finite. Applications to 3- and 4-manifolds are given:
(1) We study finite group actions on closed 4-manifolds and $\pi_1$-isomorphic cobordism of 3-dimensional lens spaces. Results including: (a) Two lens spaces are $\pi_1$-isomorphic cobordant if and only if there is a degree one map between them. (b) Each spherical 3-manifold $M\ne S^3$ can be realized as the unique non-free orbit type for a finite group action on a closed 4-manifold.
(2) The minimal bounding index $O_b(M)$ for closed 3-manifolds $M$ are defined, %and bounding Euler charicteristic $\chi_b(M)$. the relations between finiteness of $O_b(M)$ and virtual achirality of aspherical (hyperbolic) $M$ are addressed. We calculate $O_b(M)$ for some lens spaces $M$. Each prime is realized as a minimal bounding index.
(3) We also discuss some concrete examples:Surface bundle often bound surface bundles, and prime 3-manifolds often virtually bound surface bundles, $W$ bounded by some lens spaces realizing $O_b$ is constructed. | [
{
"version": "v1",
"created": "Tue, 10 Jun 2025 14:35:46 GMT"
},
{
"version": "v2",
"created": "Wed, 11 Jun 2025 13:33:11 GMT"
}
] | 2025-06-12T00:00:00 | [
[
"Lin",
"Jianfeng",
""
],
[
"Wang",
"Zhongzi",
""
]
] |
2506.08849 | Jingguo Qu | Jingguo Qu, Xinyang Han, Tonghuan Xiao, Jia Ai, Juan Wu, Tong Zhao, Jing Qin, Ann Dorothy King, Winnie Chiu-Wing Chu, Jing Cai, Michael Tin-Cheung Ying | Adapting Vision-Language Foundation Model for Next Generation Medical Ultrasound Image Analysis | null | null | null | null | cs.CV | http://creativecommons.org/licenses/by/4.0/ | Medical ultrasonography is an essential imaging technique for examining superficial organs and tissues, including lymph nodes, breast, and thyroid. It employs high-frequency ultrasound waves to generate detailed images of the internal structures of the human body. However, manually contouring regions of interest in these images is a labor-intensive task that demands expertise and often results in inconsistent interpretations among individuals. Vision-language foundation models, which have excelled in various computer vision applications, present new opportunities for enhancing ultrasound image analysis. Yet, their performance is hindered by the significant differences between natural and medical imaging domains. This research seeks to overcome these challenges by developing domain adaptation methods for vision-language foundation models. In this study, we explore the fine-tuning pipeline for vision-language foundation models by utilizing large language model as text refiner with special-designed adaptation strategies and task-driven heads. Our approach has been extensively evaluated on six ultrasound datasets and two tasks: segmentation and classification. The experimental results show that our method can effectively improve the performance of vision-language foundation models for ultrasound image analysis, and outperform the existing state-of-the-art vision-language and pure foundation models. The source code of this study is available at https://github.com/jinggqu/NextGen-UIA. | [
{
"version": "v1",
"created": "Tue, 10 Jun 2025 14:37:51 GMT"
},
{
"version": "v2",
"created": "Wed, 11 Jun 2025 03:11:14 GMT"
}
] | 2025-06-12T00:00:00 | [
[
"Qu",
"Jingguo",
""
],
[
"Han",
"Xinyang",
""
],
[
"Xiao",
"Tonghuan",
""
],
[
"Ai",
"Jia",
""
],
[
"Wu",
"Juan",
""
],
[
"Zhao",
"Tong",
""
],
[
"Qin",
"Jing",
""
],
[
"King",
"Ann Dorothy",
""
],
[
"Chu",
"Winnie Chiu-Wing",
""
],
[
"Cai",
"Jing",
""
],
[
"Ying",
"Michael Tin-Cheung",
""
]
] |
2506.08852 | Freddie Barter | Freddie Barter and Christopher R. K. Windows-Yule | Packing3D: An Open-Source Analytical Framework for Computing Packing Density and Mixing Indices Using Partial Spherical Volumes | 11 pages, 5 figures. GitHub repository for the package available at: https://github.com/fjbarter/Packing3D.jl | null | null | null | cond-mat.soft math-ph math.MP | http://creativecommons.org/licenses/by/4.0/ | Accurate quantification of local packing density and mixing in simulations of particulate systems is essential for many industrial applications. Traditional methods which simply count the number of particle centres within a given volume of space (cell) introduce discontinuities at cell boundaries, leading to unreliable measurement of packing density. We introduce Packing3D, an open-source Julia package providing analytic, partial-volume calculations for spheres intersecting Cartesian and cylindrical meshes. Our goals were to (1) eliminate boundary-artifact jumps, (2) maintain high throughput on large datasets, and (3) deliver standard mixing metrics via a unified API. We derive closed-form solutions for single, double and triple spherical-cap intersections, plus sphere-cylinder overlaps. A short-circuit bounding-sphere test shortens computations: fully inside or outside spheres are handled in $\mathcal{O}(1)$ time, and only near-boundary spheres invoke the analytic kernels. We implement efficient mesh-generation routines, principal-cell indexing, and data-splitting functions for time-series analyses. Performance and accuracy were validated against simple cubic and face-centered cubic lattices and via boundary-shift continuity tests. Packing3D converges exactly to theoretical lattice densities, eliminates discontinuities at sub-particle resolution, and processes up to $10^8$ sphere-cell intersections per second in single-threaded Julia with linear scaling in particle count. Memory usage remains modest (40 B per particle, 48 B per cell). Packing3D provides researchers with continuous, reproducible volume-fraction fields and robust mixing indices at high performance, facilitating sensitivity analyses and optimisation in granular process engineering. The package is freely available at https://github.com/fjbarter/Packing3D.jl | [
{
"version": "v1",
"created": "Tue, 10 Jun 2025 14:40:55 GMT"
},
{
"version": "v2",
"created": "Wed, 11 Jun 2025 11:01:15 GMT"
}
] | 2025-06-12T00:00:00 | [
[
"Barter",
"Freddie",
""
],
[
"Windows-Yule",
"Christopher R. K.",
""
]
] |
2506.08860 | Samah Kansab | Samah Kansab, Francis Bordeleau, Ali Tizghadam | On The Impact of Merge Request Deviations on Code Review Practices | null | null | null | null | cs.SE cs.AI cs.LG | http://creativecommons.org/licenses/by/4.0/ | Code review is a key practice in software engineering, ensuring quality and collaboration. However, industrial Merge Request (MR) workflows often deviate from standardized review processes, with many MRs serving non-review purposes (e.g., drafts, rebases, or dependency updates). We term these cases deviations and hypothesize that ignoring them biases analytics and undermines ML models for review analysis.
We identify seven deviation categories, occurring in 37.02% of MRs, and propose a few-shot learning detection method (91% accuracy). By excluding deviations, ML models predicting review completion time improve performance in 53.33% of cases (up to 2.25x) and exhibit significant shifts in feature importance (47% overall, 60% top-*k*).
Our contributions include: (1) a taxonomy of MR deviations, (2) an AI-driven detection approach, and (3) empirical evidence of their impact on ML-based review analytics. This work aids practitioners in optimizing review efforts and ensuring reliable insights. | [
{
"version": "v1",
"created": "Tue, 10 Jun 2025 14:51:20 GMT"
},
{
"version": "v2",
"created": "Wed, 11 Jun 2025 01:21:02 GMT"
}
] | 2025-06-12T00:00:00 | [
[
"Kansab",
"Samah",
""
],
[
"Bordeleau",
"Francis",
""
],
[
"Tizghadam",
"Ali",
""
]
] |
2506.08869 | Francis Valiquette | Peter J. Olver, Masoud Sabzevari, Francis Valiquette | Convergence of Normal Form Power Series for Infinite-Dimensional Lie Pseudo-Group Actions | 76 pages | null | null | null | math-ph math.DG math.MP | http://creativecommons.org/licenses/by/4.0/ | We prove the convergence of normal form power series for suitably nonsingular analytic submanifolds under a broad class of infinite-dimensional Lie pseudo-group actions. Our theorem is illustrated by a number of examples, and includes, as a particular case, Chern and Moser's celebrated convergence theorem for normal forms of real hypersurfaces. The construction of normal forms relies on the equivariant moving frame method, while the convergence proof is based on the realization that the normal form can be recovered as part of the solution to an initial value problem for an involutive system of differential equations, whose analyticity is guaranteed by the Cartan-K\"ahler Theorem. | [
{
"version": "v1",
"created": "Tue, 10 Jun 2025 15:01:37 GMT"
},
{
"version": "v2",
"created": "Wed, 11 Jun 2025 13:31:58 GMT"
}
] | 2025-06-12T00:00:00 | [
[
"Olver",
"Peter J.",
""
],
[
"Sabzevari",
"Masoud",
""
],
[
"Valiquette",
"Francis",
""
]
] |
2506.08877 | Krzysztof Ptaszynski | Krzysztof Ptaszynski, Timur Aslyamov, Massimiliano Esposito | Nonequilibrium fluctuation-response relations for state-current correlations | 10 pages, 6 figures. Companion paper to arXiv:2412.10233 | null | null | null | cond-mat.stat-mech | http://creativecommons.org/licenses/by/4.0/ | Recently, novel exact identities known as Fluctuation-Response Relations (FRRs) have been derived for nonequilibrium steady states of Markov jump processes. These identities link the fluctuations of state or current observables to a combination of responses of these observables to perturbations of transition rates. Here, we complement these results by deriving analogous FRRs applicable to mixed covariances of one state and one current observable. We further derive novel Inverse FRRs expressing individual state or current response in terms of a combination of covariances rather than vice versa. Using these relations, we demonstrate that the breaking of the Onsager symmetry can occur only in the presence of state-current correlations. On the practical side, we demonstrate the applicability of FRRs for simplifying calculations of fluctuations in large Markov networks, we use them to explain the behavior of fluctuations in quantum dot devices or enzymatic reaction schemes, and discuss their potential relevance for model inference. | [
{
"version": "v1",
"created": "Tue, 10 Jun 2025 15:07:11 GMT"
},
{
"version": "v2",
"created": "Wed, 11 Jun 2025 14:54:08 GMT"
}
] | 2025-06-12T00:00:00 | [
[
"Ptaszynski",
"Krzysztof",
""
],
[
"Aslyamov",
"Timur",
""
],
[
"Esposito",
"Massimiliano",
""
]
] |
2506.08885 | Vasu Sharma | Danush Khanna, Krishna Kumar, Basab Ghosh, Vinija Jain, Vasu Sharma, Aman Chadha, Amitava Das | AdversariaL attacK sAfety aLIgnment(ALKALI): Safeguarding LLMs through GRACE: Geometric Representation-Aware Contrastive Enhancement- Introducing Adversarial Vulnerability Quality Index (AVQI) | null | null | null | null | cs.CL cs.LG | http://creativecommons.org/licenses/by/4.0/ | Adversarial threats against LLMs are escalating faster than current defenses can adapt. We expose a critical geometric blind spot in alignment: adversarial prompts exploit latent camouflage, embedding perilously close to the safe representation manifold while encoding unsafe intent thereby evading surface level defenses like Direct Preference Optimization (DPO), which remain blind to the latent geometry. We introduce ALKALI, the first rigorously curated adversarial benchmark and the most comprehensive to date spanning 9,000 prompts across three macro categories, six subtypes, and fifteen attack families. Evaluation of 21 leading LLMs reveals alarmingly high Attack Success Rates (ASRs) across both open and closed source models, exposing an underlying vulnerability we term latent camouflage, a structural blind spot where adversarial completions mimic the latent geometry of safe ones. To mitigate this vulnerability, we introduce GRACE - Geometric Representation Aware Contrastive Enhancement, an alignment framework coupling preference learning with latent space regularization. GRACE enforces two constraints: latent separation between safe and adversarial completions, and adversarial cohesion among unsafe and jailbreak behaviors. These operate over layerwise pooled embeddings guided by a learned attention profile, reshaping internal geometry without modifying the base model, and achieve up to 39% ASR reduction. Moreover, we introduce AVQI, a geometry aware metric that quantifies latent alignment failure via cluster separation and compactness. AVQI reveals when unsafe completions mimic the geometry of safe ones, offering a principled lens into how models internally encode safety. We make the code publicly available at https://anonymous.4open.science/r/alkali-B416/README.md. | [
{
"version": "v1",
"created": "Tue, 10 Jun 2025 15:14:17 GMT"
},
{
"version": "v2",
"created": "Wed, 11 Jun 2025 05:27:54 GMT"
}
] | 2025-06-12T00:00:00 | [
[
"Khanna",
"Danush",
""
],
[
"Kumar",
"Krishna",
""
],
[
"Ghosh",
"Basab",
""
],
[
"Jain",
"Vinija",
""
],
[
"Sharma",
"Vasu",
""
],
[
"Chadha",
"Aman",
""
],
[
"Das",
"Amitava",
""
]
] |
2506.08895 | Lucie Cros | Lucie Cros, Antoine Gusdorf, Philippe Salom\'e, Sergiy Stepkin, Philippe Zarka, Pedro Salas, Alan Loh, Pierre Lesaffre, Jonathan Freundlich, Marta Alves, Fran\c{c}ois Boulanger, Andrea Bracco, St\'ephane Corbel, Maryvonne Gerin, Javier Goicoechea, Isabelle Grenier, Jean-Mathias Grie{\ss}meier, Martin Houde, Oleksandr Konovalenko, Antoine Marchal, Alexandre Marcowith, Florent Mertens, Fr\'ed\'erique Motte, Michel Tagger, Alexander Tielens, Gilles Theureau, Peter Tokarsky, Oleg Ulyanov, Vyacheslav Zakharenko | Observations of Carbon Radio Recombination Lines with the NenuFAR telescope. I. Cassiopeia A and Cygnus A | null | null | null | null | astro-ph.GA | http://creativecommons.org/licenses/by/4.0/ | Carbon Radio Recombination Lines (CRRLs) at decametre wavelengths trace the diffuse phase of the interstellar medium (ISM) of the Galaxy. Their observation allows to measure physical parameters of this phase. We observed CRRLs with the recently commissioned New Extension in Nan\c{c}ay Upgrading LOFAR (NenuFAR) telescope towards two of the brightest sources at low-frequency (10-85 MHz): Cassiopeia A and Cygnus A (hereafter Cas A and Cyg A respectively), to measure the density n_e and temperature T_e of electrons in line-of-sight clouds. We used NenuFAR's beamforming mode, and we integrated several tens of hours on each source. The nominal spectral resolution was 95.4 Hz. We developed a pipeline to remove radio frequency interference (RFI) contamination and correct the baselines. We then fitted the spectral lines observed in absorption, associated to line-of-sight clouds. Cas A is the brightest source in the sky at low frequencies and represents an appropriate test bench for this new telescope. On this source, we detected 398 C\alpha lines between principal quantum numbers n=426 and n=826. C\alpha lines towards Cyg A were fainter. We stacked the signal by groups of a few tens of lines to improve the quality of our fitting process. On both sources we reached significantly higher S/N and spectral resolution than the most recent detections by the LOw Frequency ARray (LOFAR). The variation of line shape with n provides constraints on the physical properties of the clouds: T_e, n_e, the temperature T_0 of the radiation field, the mean turbulent velocity v_t and the typical size of the cloud. The NenuFAR observations sample a larger space volume than LOFAR's towards the same sources due to the differences in instrumental beamsizes, and the discrepancies highlight the sensitivity of low-frequency CRRLs as probes of the diffuse ISM, paving the way towards large area surveys of CRRLs in our Galaxy. | [
{
"version": "v1",
"created": "Tue, 10 Jun 2025 15:22:39 GMT"
},
{
"version": "v2",
"created": "Wed, 11 Jun 2025 12:32:57 GMT"
}
] | 2025-06-12T00:00:00 | [
[
"Cros",
"Lucie",
""
],
[
"Gusdorf",
"Antoine",
""
],
[
"Salomé",
"Philippe",
""
],
[
"Stepkin",
"Sergiy",
""
],
[
"Zarka",
"Philippe",
""
],
[
"Salas",
"Pedro",
""
],
[
"Loh",
"Alan",
""
],
[
"Lesaffre",
"Pierre",
""
],
[
"Freundlich",
"Jonathan",
""
],
[
"Alves",
"Marta",
""
],
[
"Boulanger",
"François",
""
],
[
"Bracco",
"Andrea",
""
],
[
"Corbel",
"Stéphane",
""
],
[
"Gerin",
"Maryvonne",
""
],
[
"Goicoechea",
"Javier",
""
],
[
"Grenier",
"Isabelle",
""
],
[
"Grießmeier",
"Jean-Mathias",
""
],
[
"Houde",
"Martin",
""
],
[
"Konovalenko",
"Oleksandr",
""
],
[
"Marchal",
"Antoine",
""
],
[
"Marcowith",
"Alexandre",
""
],
[
"Mertens",
"Florent",
""
],
[
"Motte",
"Frédérique",
""
],
[
"Tagger",
"Michel",
""
],
[
"Tielens",
"Alexander",
""
],
[
"Theureau",
"Gilles",
""
],
[
"Tokarsky",
"Peter",
""
],
[
"Ulyanov",
"Oleg",
""
],
[
"Zakharenko",
"Vyacheslav",
""
]
] |
2506.08900 | Jos\'e Morano | Jos\'e Morano, Botond Fazekas, Emese S\"ukei, Ronald Fecso, Taha Emre, Markus Gumpinger, Georg Faustmann, Marzieh Oghbaie, Ursula Schmidt-Erfurth, Hrvoje Bogunovi\'c | MIRAGE: Multimodal foundation model and benchmark for comprehensive retinal OCT image analysis | null | null | null | null | cs.CV | http://creativecommons.org/licenses/by-nc-nd/4.0/ | Artificial intelligence (AI) has become a fundamental tool for assisting clinicians in analyzing ophthalmic images, such as optical coherence tomography (OCT). However, developing AI models often requires extensive annotation, and existing models tend to underperform on independent, unseen data. Foundation models (FMs), large AI models trained on vast unlabeled datasets, have shown promise in overcoming these challenges. Nonetheless, available FMs for ophthalmology lack extensive validation, especially for segmentation tasks, and focus on a single imaging modality. In this context, we propose MIRAGE, a novel multimodal FM for the analysis of OCT and scanning laser ophthalmoscopy (SLO) images. Additionally, we propose a new evaluation benchmark with OCT/SLO classification and segmentation tasks. The comparison with general and specialized FMs and segmentation methods shows the superiority of MIRAGE in both types of tasks, highlighting its suitability as a basis for the development of robust AI systems for retinal OCT image analysis. Both MIRAGE and the evaluation benchmark are publicly available: https://github.com/j-morano/MIRAGE. | [
{
"version": "v1",
"created": "Tue, 10 Jun 2025 15:25:55 GMT"
},
{
"version": "v2",
"created": "Wed, 11 Jun 2025 09:38:22 GMT"
}
] | 2025-06-12T00:00:00 | [
[
"Morano",
"José",
""
],
[
"Fazekas",
"Botond",
""
],
[
"Sükei",
"Emese",
""
],
[
"Fecso",
"Ronald",
""
],
[
"Emre",
"Taha",
""
],
[
"Gumpinger",
"Markus",
""
],
[
"Faustmann",
"Georg",
""
],
[
"Oghbaie",
"Marzieh",
""
],
[
"Schmidt-Erfurth",
"Ursula",
""
],
[
"Bogunović",
"Hrvoje",
""
]
] |
2506.08908 | Jiajun Li | Jiajun Li (1 and 5), Yue Ma (2), Xinyu Zhang (1), Qingyan Wei (3), Songhua Liu (4 and 5), Linfeng Zhang (5) ((1) University of Electronic Science and Technology of China, (2) The Hong Kong University of Science and Technology, (3) Central South University, (4) National University of Singapore, (5) Shanghai Jiaotong University) | SkipVAR: Accelerating Visual Autoregressive Modeling via Adaptive Frequency-Aware Skipping | null | null | null | null | cs.CV | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Recent studies on Visual Autoregressive (VAR) models have highlighted that high-frequency components, or later steps, in the generation process contribute disproportionately to inference latency. However, the underlying computational redundancy involved in these steps has yet to be thoroughly investigated. In this paper, we conduct an in-depth analysis of the VAR inference process and identify two primary sources of inefficiency: step redundancy and unconditional branch redundancy. To address step redundancy, we propose an automatic step-skipping strategy that selectively omits unnecessary generation steps to improve efficiency. For unconditional branch redundancy, we observe that the information gap between the conditional and unconditional branches is minimal. Leveraging this insight, we introduce unconditional branch replacement, a technique that bypasses the unconditional branch to reduce computational cost. Notably, we observe that the effectiveness of acceleration strategies varies significantly across different samples. Motivated by this, we propose SkipVAR, a sample-adaptive framework that leverages frequency information to dynamically select the most suitable acceleration strategy for each instance. To evaluate the role of high-frequency information, we introduce high-variation benchmark datasets that test model sensitivity to fine details. Extensive experiments show SkipVAR achieves over 0.88 average SSIM with up to 1.81x overall acceleration and 2.62x speedup on the GenEval benchmark, maintaining model quality. These results confirm the effectiveness of frequency-aware, training-free adaptive acceleration for scalable autoregressive image generation. Our code is available at https://github.com/fakerone-li/SkipVAR and has been publicly released. | [
{
"version": "v1",
"created": "Tue, 10 Jun 2025 15:35:29 GMT"
},
{
"version": "v2",
"created": "Wed, 11 Jun 2025 04:58:42 GMT"
}
] | 2025-06-12T00:00:00 | [
[
"Li",
"Jiajun",
"",
"1 and 5"
],
[
"Ma",
"Yue",
"",
"The Hong Kong University of Science and Technology"
],
[
"Zhang",
"Xinyu",
"",
"University of Electronic Science and Technology of China"
],
[
"Wei",
"Qingyan",
"",
"Central South University"
],
[
"Liu",
"Songhua",
"",
"4 and 5"
],
[
"Zhang",
"Linfeng",
"",
"Shanghai Jiaotong University"
]
] |
2506.08913 | V. T. Renard | F. Mesple and P. Mallet and G. Trambly de Laissardi\`ere and C. Dutreix and G. Lapertot and J-Y. Veuillen and V. T. Renard | Experimental evidence of the topological obstruction in twisted bilayer graphene | Added one missing author (GTL). 5 pages, 3 figures, supplements on request | null | null | null | cond-mat.mes-hall | http://creativecommons.org/licenses/by-nc-nd/4.0/ | The rich physics of magic angle twisted bilayer graphene (TBG) results from the Coulomb interactions of electrons in flat bands of non-trivial topology. While the bands' dispersion is well characterized, accessing their topology remains an experimental challenge. Recent measurements established the local density of states (LDOS) as a topological observable. Here, we use scanning tunnelling microscopy to investigate the LDOS of TBG near a defect. We observe characteristic patterns resulting from the Dirac cones having the same chirality within a moir\'e valley. At higher energies, we observe the Lifshitz transition associated with the Dirac cones mixing. Our measurements provide a full characterization of TBG's band structure, confirming the main features of the continuum model including the renormalization of the Fermi velocity, the role of emergent symmetries and the topological obstruction of the wavefunctions. | [
{
"version": "v1",
"created": "Tue, 10 Jun 2025 15:39:45 GMT"
},
{
"version": "v2",
"created": "Wed, 11 Jun 2025 06:43:22 GMT"
}
] | 2025-06-12T00:00:00 | [
[
"Mesple",
"F.",
""
],
[
"Mallet",
"P.",
""
],
[
"de Laissardière",
"G. Trambly",
""
],
[
"Dutreix",
"C.",
""
],
[
"Lapertot",
"G.",
""
],
[
"Veuillen",
"J-Y.",
""
],
[
"Renard",
"V. T.",
""
]
] |
2506.08934 | Ryoko Oishi-Tomiyasu Dr. | Ryoko Oishi-Tomiyasu | Piecewise-linear embeddings of the space of 3D lattices into $\RR^{13}$ for high-throughput handling of lattice parameters | null | null | null | null | math.MG | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | We present two methods to continuously and piecewise-linearly parametrize rank-3 lattices by vectors of $\RR^{13}$, which provides an efficient way to judge if two sets of parameters provide nearly identical lattices within their margins of errors. Such a parametrization can be used to speed up scientific computing involving periodic structures in $\RR^3$ such as crystal structures, which includes database querying, detection of duplicate entries, and structure generation via deep learning techniques. One gives a novel application of Conway's vonorms and conorms, and another is achieved through a natural extension of Ry{\u s}hkov's $C$-type to the setting modulo $3$. Voronoi vectors modulo 3 obtained in the latter approach provide an algorithm for enumerating of all potential isometries under perturbations of lattice parameters. | [
{
"version": "v1",
"created": "Tue, 10 Jun 2025 15:59:49 GMT"
},
{
"version": "v2",
"created": "Wed, 11 Jun 2025 14:43:32 GMT"
}
] | 2025-06-12T00:00:00 | [
[
"Oishi-Tomiyasu",
"Ryoko",
""
]
] |
2506.08942 | Maur\'icio Corr\^ea | Omegar Calvo-Andrade, Maur\'icio Corr\^ea, Marcos Jardim, Jos\'e Seade | On the connectedness of the singular set of holomorphic foliations | null | null | null | null | math.AG math.CV math.DG math.DS | http://creativecommons.org/licenses/by/4.0/ | Let $\mathcal{F}$ be a codimension 1 singular holomorphic foliation on $\mathbb{P}^3$ and let $Z$ be the union of the 1-dimensional connected components of its singular set; Dominique Cerveau asked in 2013 whether $Z$ is necessarily connected. We prove that if a $k$-dimensional holomorphic foliation on a projective manifold whose determinant bundle of the normal sheaf is ample has a singular set $Z$ of dimension $k-1$, then $Z$ is connected. This can be regarded as providing an obstruction for the integrability of singular holomorphic distributions in the spirit of Bott's work. | [
{
"version": "v1",
"created": "Tue, 10 Jun 2025 16:05:15 GMT"
},
{
"version": "v2",
"created": "Wed, 11 Jun 2025 09:57:45 GMT"
}
] | 2025-06-12T00:00:00 | [
[
"Calvo-Andrade",
"Omegar",
""
],
[
"Corrêa",
"Maurício",
""
],
[
"Jardim",
"Marcos",
""
],
[
"Seade",
"José",
""
]
] |
2506.08952 | Clara Lachenmaier | Clara Lachenmaier and Judith Sieker and Sina Zarrie{\ss} | Can LLMs Ground when they (Don't) Know: A Study on Direct and Loaded Political Questions | Preprint accepted at ACL Main Conference 2025 | null | null | null | cs.CL cs.AI | http://creativecommons.org/licenses/by/4.0/ | Communication among humans relies on conversational grounding, allowing interlocutors to reach mutual understanding even when they do not have perfect knowledge and must resolve discrepancies in each other's beliefs. This paper investigates how large language models (LLMs) manage common ground in cases where they (don't) possess knowledge, focusing on facts in the political domain where the risk of misinformation and grounding failure is high. We examine the ability of LLMs to answer direct knowledge questions and loaded questions that presuppose misinformation. We evaluate whether loaded questions lead LLMs to engage in active grounding and correct false user beliefs, in connection to their level of knowledge and their political bias. Our findings highlight significant challenges in LLMs' ability to engage in grounding and reject false user beliefs, raising concerns about their role in mitigating misinformation in political discourse. | [
{
"version": "v1",
"created": "Tue, 10 Jun 2025 16:20:09 GMT"
},
{
"version": "v2",
"created": "Wed, 11 Jun 2025 06:58:55 GMT"
}
] | 2025-06-12T00:00:00 | [
[
"Lachenmaier",
"Clara",
""
],
[
"Sieker",
"Judith",
""
],
[
"Zarrieß",
"Sina",
""
]
] |
2506.08960 | Tao-Ran Hu | Tao-Ran Hu, Katsuyoshi Sone, Feng-Kun Guo, Tetsuo Hyodo, Ian Low | Entanglement Suppression, Quantum Statistics and Symmetries in Spin-3/2 Baryon Scatterings | 38 pages, 1 figure | null | null | null | hep-ph hep-th nucl-th quant-ph | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | We explore the interplay among entanglement suppression, quantum statistics and enhanced symmetries in the non-relativistic $S$-wave scattering involving the lowest-lying spin-3/2 baryons, which can be considered as four-dimensional qudits. These baryons form a ten-dimensional representation (decuplet) under the $\text{SU}(3)$ light-flavor symmetry and, in this limit, are considered indistinguishable under strong interactions. Treating the $S$-matrix in the spin-3/2 baryon-baryon scattering as a quantum logic gate in the spin space, we study the consequence of entanglement suppression and compute the entanglement power of the $S$-matrix. When the entanglement power vanishes, the $S$-matrix is either an Identity or a SWAP gate and spin-flavor symmetries and/or non-relativistic conformal invariance emerge, as previously observed in spin-1/2 baryons. In the case of scattering identical particles, the entanglement power never vanishes due to constraints from spin statistics, which we interpret as projection-valued measurements onto symmetric or antisymmetric Hilbert space and define the entanglement power accordingly. When the entanglement power is non-vanishing but sits at a global or local minimum, enhanced symmetries still emerge and the $S$-matrix can be interpreted as an Identity or a SWAP gate acting on the restricted Hilbert space allowed by quantum statistics. In general, when scattering identical spin-$s$ particles, we identify an enhanced $\text{SU}(2s+1)_{\text{spin}}$ symmetry for the Identity gate. | [
{
"version": "v1",
"created": "Tue, 10 Jun 2025 16:31:52 GMT"
},
{
"version": "v2",
"created": "Wed, 11 Jun 2025 13:43:06 GMT"
}
] | 2025-06-12T00:00:00 | [
[
"Hu",
"Tao-Ran",
""
],
[
"Sone",
"Katsuyoshi",
""
],
[
"Guo",
"Feng-Kun",
""
],
[
"Hyodo",
"Tetsuo",
""
],
[
"Low",
"Ian",
""
]
] |
2506.08976 | Yu Wang | Yu Wang, Shuyuan Xu, Xueda Wei, Xinrui Luo, Stephen Shing-Toung Yau, Shing-Tung Yau, Rongling Wu | Yau-YauAL: A computer tool for solving nonlinear filtering problems | 8 pages, 5 figures | null | null | null | stat.AP | http://creativecommons.org/licenses/by/4.0/ | The Yau-Yau nonlinear filter has increasingly emerged as a powerful tool to study stochastic complex systems. To leverage it to a wider spectrum of application scenarios, we pack the Yau-Yau filtering ALgorithms (YauYauAL) into a package of computer software. Yau-YauAL was written in R, designed to simplify the implementation of the Yau-Yau filter for solving nonlinear filtering problems. Combining R's accessibility with C++ (via Rcpp) for computational efficiency, YauYauAL provides an intuitive Shiny-based interface that enables real-time parameter adjustment and result visualization. At its core, YauYauAL employs finite difference methods to numerically solve the Kolmogorov forward equation, ensuring a stable and accurate solution even for complex systems. YauYauAL's modular design and open-source framework further encourage customization and community-driven development. YauYauAL aims to bridge the gap between theoretical nonlinear filtering methods and practical applications, without requiring expertise in differential equation solving or programming, fostering its broader impact on various scientific fields, such as signal processing, finance, medicine, and biology among a long list. | [
{
"version": "v1",
"created": "Tue, 10 Jun 2025 16:45:54 GMT"
},
{
"version": "v2",
"created": "Wed, 11 Jun 2025 06:07:36 GMT"
}
] | 2025-06-12T00:00:00 | [
[
"Wang",
"Yu",
""
],
[
"Xu",
"Shuyuan",
""
],
[
"Wei",
"Xueda",
""
],
[
"Luo",
"Xinrui",
""
],
[
"Yau",
"Stephen Shing-Toung",
""
],
[
"Yau",
"Shing-Tung",
""
],
[
"Wu",
"Rongling",
""
]
] |
2506.08980 | He Kaifeng | Kaifeng He, Mingwei Liu, Chong Wang, Zike Li, Yanlin Wang, Xin Peng, Zibin Zheng | Towards Better Code Generation: Adaptive Decoding with Uncertainty Guidance | null | null | null | null | cs.SE | http://creativecommons.org/licenses/by/4.0/ | Code generation using large language models (LLMs) is highly sensitive to the choice of tokens during decoding, especially at points of uncertainty that critically affect the generated program's logic. Conventional decoding methods such as greedy search and beam search apply uniform treatment to all tokens, neglecting the unique uncertainty characteristics inherent in code generation, which can result in suboptimal outputs. In this work, we conduct an empirical analysis demonstrating that a significant portion of generation errors arises from incorrect token ranking at high-uncertainty steps, where the ground truth token exists in the candidate set but fails to be ranked first.
Inspired by this insight, we introduce AdaDec, an adaptive decoding framework guided by token-level uncertainty quantified via Shannon entropy. AdaDec dynamically learns uncertainty thresholds tailored to each model and employs a pause-then-rerank mechanism with lookahead when the uncertainty surpasses these thresholds. Evaluation on the HumanEval and MBPP benchmarks reveals that AdaDec achieves up to a 15.5% improvement in Pass@1 accuracy compared to greedy decoding, matches or outperforms traditional beam search, and reduces both computational overhead and latency through targeted, selective pausing. Our findings suggest that uncertainty-aware adaptive decoding holds considerable potential for enhancing both the reliability and efficiency of code generation with LLMs. | [
{
"version": "v1",
"created": "Tue, 10 Jun 2025 16:49:46 GMT"
},
{
"version": "v2",
"created": "Wed, 11 Jun 2025 04:29:51 GMT"
}
] | 2025-06-12T00:00:00 | [
[
"He",
"Kaifeng",
""
],
[
"Liu",
"Mingwei",
""
],
[
"Wang",
"Chong",
""
],
[
"Li",
"Zike",
""
],
[
"Wang",
"Yanlin",
""
],
[
"Peng",
"Xin",
""
],
[
"Zheng",
"Zibin",
""
]
] |
2506.08982 | Ivan Rubachev | Ivan Rubachev, Akim Kotelnikov, Nikolay Kartashev, Artem Babenko | On Finetuning Tabular Foundation Models | null | null | null | null | cs.LG | http://creativecommons.org/licenses/by-nc-sa/4.0/ | Foundation models are an emerging research direction in tabular deep learning. Notably, TabPFNv2 recently claimed superior performance over traditional GBDT-based methods on small-scale datasets using an in-context learning paradigm, which does not adapt model parameters to target datasets. However, the optimal finetuning approach for adapting tabular foundational models, and how this adaptation reshapes their internal mechanisms, remains underexplored. While prior works studied finetuning for earlier foundational models, inconsistent findings and TabPFNv2's unique architecture necessitate fresh investigation. To address these questions, we first systematically evaluate various finetuning strategies on diverse datasets. Our findings establish full finetuning as the most practical solution for TabPFNv2 in terms of time-efficiency and effectiveness. We then investigate how finetuning alters TabPFNv2's inner mechanisms, drawing an analogy to retrieval-augmented models. We reveal that the success of finetuning stems from the fact that after gradient-based adaptation, the dot products of the query-representations of test objects and the key-representations of in-context training objects more accurately reflect their target similarity. This improved similarity allows finetuned TabPFNv2 to better approximate target dependency by appropriately weighting relevant in-context samples, improving the retrieval-based prediction logic. From the practical perspective, we managed to finetune TabPFNv2 on datasets with up to 50K objects, observing performance improvements on almost all tasks. More precisely, on academic datasets with I.I.D. splits, finetuning allows TabPFNv2 to achieve state-of-the-art results, while on datasets with gradual temporal shifts and rich feature sets, TabPFNv2 is less stable and prior methods remain better. | [
{
"version": "v1",
"created": "Tue, 10 Jun 2025 16:52:31 GMT"
},
{
"version": "v2",
"created": "Wed, 11 Jun 2025 09:47:49 GMT"
}
] | 2025-06-12T00:00:00 | [
[
"Rubachev",
"Ivan",
""
],
[
"Kotelnikov",
"Akim",
""
],
[
"Kartashev",
"Nikolay",
""
],
[
"Babenko",
"Artem",
""
]
] |
2506.09002 | Bei Chu | Bei Chu, Yang Feng, Kui Liu, Hange Shi, Zifan Nan, Zhaoqiang Guo, Baowen Xu | Boosting Rust Unit Test Coverage through Hybrid Program Analysis and Large Language Models | 10 pages, 5 figures | null | null | null | cs.SE | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Unit testing is essential for ensuring software reliability and correctness. Classic Search-Based Software Testing (SBST) methods and concolic execution-based approaches for generating unit tests often fail to achieve high coverage due to difficulties in handling complex program units, such as branching conditions and external dependencies. Recent work has increasingly utilized large language models (LLMs) to generate test cases, improving the quality of test generation by providing better context and correcting errors in the model's output. However, these methods rely on fixed prompts, resulting in relatively low compilation success rates and coverage. This paper presents PALM, an approach that leverages large language models (LLMs) to enhance the generation of high-coverage unit tests. PALM performs program analysis to identify branching conditions within functions, which are then combined into path constraints. These constraints and relevant contextual information are used to construct prompts that guide the LLMs in generating unit tests. We implement the approach and evaluate it in 10 open-source Rust crates. Experimental results show that within just two or three hours, PALM can significantly improves test coverage compared to classic methods, with increases in overall project coverage exceeding 50% in some instances and its generated tests achieving an average coverage of 75.77%, comparable to human effort (71.30%), highlighting the potential of LLMs in automated test generation. We submitted 91 PALM-generated unit tests targeting new code. Of these submissions, 80 were accepted, 5 were rejected, and 6 remain pending review. The results demonstrate the effectiveness of integrating program analysis with AI and open new avenues for future research in automated software testing. | [
{
"version": "v1",
"created": "Tue, 10 Jun 2025 17:21:21 GMT"
},
{
"version": "v2",
"created": "Wed, 11 Jun 2025 03:16:55 GMT"
}
] | 2025-06-12T00:00:00 | [
[
"Chu",
"Bei",
""
],
[
"Feng",
"Yang",
""
],
[
"Liu",
"Kui",
""
],
[
"Shi",
"Hange",
""
],
[
"Nan",
"Zifan",
""
],
[
"Guo",
"Zhaoqiang",
""
],
[
"Xu",
"Baowen",
""
]
] |
2506.09003 | Lei Zhang | Lei Zhang, Jiaxi Yang, Min Yang, Jian Yang, Mouxiang Chen, Jiajun Zhang, Zeyu Cui, Binyuan Hui, Junyang Lin | SWE-Flow: Synthesizing Software Engineering Data in a Test-Driven Manner | Accepted by ICML2025 | null | null | null | cs.CL | http://creativecommons.org/licenses/by-nc-nd/4.0/ | We introduce **SWE-Flow**, a novel data synthesis framework grounded in Test-Driven Development (TDD). Unlike existing software engineering data that rely on human-submitted issues, **SWE-Flow** automatically infers incremental development steps directly from unit tests, which inherently encapsulate high-level requirements. The core of **SWE-Flow** is the construction of a Runtime Dependency Graph (RDG), which precisely captures function interactions, enabling the generation of a structured, step-by-step *development schedule*. At each step, **SWE-Flow** produces a partial codebase, the corresponding unit tests, and the necessary code modifications, resulting in fully verifiable TDD tasks. With this approach, we generated 16,061 training instances and 2,020 test instances from real-world GitHub projects, creating the **SWE-Flow-Eval** benchmark. Our experiments show that fine-tuning open model on this dataset significantly improves performance in TDD-based coding. To facilitate further research, we release all code, datasets, models, and Docker images at [Github](https://github.com/Hambaobao/SWE-Flow). | [
{
"version": "v1",
"created": "Tue, 10 Jun 2025 17:23:33 GMT"
},
{
"version": "v2",
"created": "Wed, 11 Jun 2025 03:30:10 GMT"
}
] | 2025-06-12T00:00:00 | [
[
"Zhang",
"Lei",
""
],
[
"Yang",
"Jiaxi",
""
],
[
"Yang",
"Min",
""
],
[
"Yang",
"Jian",
""
],
[
"Chen",
"Mouxiang",
""
],
[
"Zhang",
"Jiajun",
""
],
[
"Cui",
"Zeyu",
""
],
[
"Hui",
"Binyuan",
""
],
[
"Lin",
"Junyang",
""
]
] |
2506.09007 | Pranam Chatterjee | Sophia Tang, Yinuo Zhang, Alexander Tong, Pranam Chatterjee | Branched Schr\"odinger Bridge Matching | null | null | null | null | cs.LG q-bio.QM | http://creativecommons.org/licenses/by-nc-nd/4.0/ | Predicting the intermediate trajectories between an initial and target distribution is a central problem in generative modeling. Existing approaches, such as flow matching and Schr\"odinger Bridge Matching, effectively learn mappings between two distributions by modeling a single stochastic path. However, these methods are inherently limited to unimodal transitions and cannot capture branched or divergent evolution from a common origin to multiple distinct outcomes. To address this, we introduce Branched Schr\"odinger Bridge Matching (BranchSBM), a novel framework that learns branched Schr\"odinger bridges. BranchSBM parameterizes multiple time-dependent velocity fields and growth processes, enabling the representation of population-level divergence into multiple terminal distributions. We show that BranchSBM is not only more expressive but also essential for tasks involving multi-path surface navigation, modeling cell fate bifurcations from homogeneous progenitor states, and simulating diverging cellular responses to perturbations. | [
{
"version": "v1",
"created": "Tue, 10 Jun 2025 17:29:48 GMT"
}
] | 2025-06-12T00:00:00 | [
[
"Tang",
"Sophia",
""
],
[
"Zhang",
"Yinuo",
""
],
[
"Tong",
"Alexander",
""
],
[
"Chatterjee",
"Pranam",
""
]
] |
2506.09009 | Hakyung Sung | Hakyung Sung, Gyu-Ho Shin, Chanyoung Lee, You Kyung Sung, Boo Kyung Jung | UD-KSL Treebank v1.3: A semi-automated framework for aligning XPOS-extracted units with UPOS tags | null | null | null | null | cs.CL | http://creativecommons.org/licenses/by-nc-sa/4.0/ | The present study extends recent work on Universal Dependencies annotations for second-language (L2) Korean by introducing a semi-automated framework that identifies morphosyntactic constructions from XPOS sequences and aligns those constructions with corresponding UPOS categories. We also broaden the existing L2-Korean corpus by annotating 2,998 new sentences from argumentative essays. To evaluate the impact of XPOS-UPOS alignments, we fine-tune L2-Korean morphosyntactic analysis models on datasets both with and without these alignments, using two NLP toolkits. Our results indicate that the aligned dataset not only improves consistency across annotation layers but also enhances morphosyntactic tagging and dependency-parsing accuracy, particularly in cases of limited annotated data. | [
{
"version": "v1",
"created": "Tue, 10 Jun 2025 17:37:56 GMT"
},
{
"version": "v2",
"created": "Wed, 11 Jun 2025 05:35:33 GMT"
}
] | 2025-06-12T00:00:00 | [
[
"Sung",
"Hakyung",
""
],
[
"Shin",
"Gyu-Ho",
""
],
[
"Lee",
"Chanyoung",
""
],
[
"Sung",
"You Kyung",
""
],
[
"Jung",
"Boo Kyung",
""
]
] |
2506.09012 | David Berger | David Berger and Alexander Lindner | Invertible Complex Measures on Euclidean Spaces | null | null | null | null | math.PR math.FA | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | In 1971 Taylor characterised all complex measures on $\mathbb{R}$ that are invertible with respect to convolution as those which can be written in the form $\delta_\gamma \ast \sigma^{\ast m} \ast \exp(\nu)$ for some $\gamma\in \mathbb{R}$, some complex measure $\nu$, some $m\in \mathbb{Z}$ and a given fixed invertible finite signed measure $\sigma$ (which has characteristic function $\mathbb{R} \ni z \mapsto (1+i z)/(1-i z)$). We extend Taylor's result to complex measures on $\mathbb{R}^n$. Somewhat surprisingly, the structure of invertible complex measures on $\mathbb{R}^n$ is not much more complicated than that of complex measures on $\mathbb{R}$, in the sense that they can be represented as $\delta_\gamma \ast \sigma_1^{\ast m_1} \ast \ldots \ast \sigma_p^{\ast m_p} \ast \exp(\nu)$ for some $\gamma \in \mathbb{R}^n$, some complex measure $\nu$ and $m_1,\ldots, m_p\in \mathbb{Z}$, where the $\sigma_i$ correspond to $\sigma$ in the one-dimensional case and actually live on $1$-dimensional subspaces of $\mathbb{R}^n$. Our proof relies on a general result of Taylor for invertible complex measures on locally compact abelian groups. To apply Taylor's result, we extend some existing results for $\mathbb{C}$-valued functions to functions with values in a semisimple commutative unital Banach algebra with connected Gelfand space. The study of invertible complex measures on $\mathbb{R}^n$ has some impact on the theory of quasi-infinitely divisible probability distributions on $\mathbb{R}^n$. | [
{
"version": "v1",
"created": "Tue, 10 Jun 2025 17:41:40 GMT"
},
{
"version": "v2",
"created": "Wed, 11 Jun 2025 10:12:48 GMT"
}
] | 2025-06-12T00:00:00 | [
[
"Berger",
"David",
""
],
[
"Lindner",
"Alexander",
""
]
] |
2506.09019 | Joel Castillo-Rey | Joel Castillo-Rey | Strong Watanabe-Yoshida conjecture for Complete Intersections | 35 pages | null | null | null | math.AC math.AG | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | In this paper, we prove the strong form of the Watanabe-Yoshida conjecture for complete intersection singularities in every positive characteristic. In characteristics 2 and 3, we explicitly compute the Hilbert-Kunz functions of the A1 and A2 singularities. | [
{
"version": "v1",
"created": "Tue, 10 Jun 2025 17:44:34 GMT"
},
{
"version": "v2",
"created": "Wed, 11 Jun 2025 16:47:28 GMT"
}
] | 2025-06-12T00:00:00 | [
[
"Castillo-Rey",
"Joel",
""
]
] |
2506.09021 | Hakyung Sung | Hakyung Sung, Karla Csuros, Min-Chang Sung | Comparing human and LLM proofreading in L2 writing: Impact on lexical and syntactic features | null | null | null | null | cs.CL | http://creativecommons.org/licenses/by-nc-sa/4.0/ | This study examines the lexical and syntactic interventions of human and LLM proofreading aimed at improving overall intelligibility in identical second language writings, and evaluates the consistency of outcomes across three LLMs (ChatGPT-4o, Llama3.1-8b, Deepseek-r1-8b). Findings show that both human and LLM proofreading enhance bigram lexical features, which may contribute to better coherence and contextual connectedness between adjacent words. However, LLM proofreading exhibits a more generative approach, extensively reworking vocabulary and sentence structures, such as employing more diverse and sophisticated vocabulary and incorporating a greater number of adjective modifiers in noun phrases. The proofreading outcomes are highly consistent in major lexical and syntactic features across the three models. | [
{
"version": "v1",
"created": "Tue, 10 Jun 2025 17:49:10 GMT"
},
{
"version": "v2",
"created": "Wed, 11 Jun 2025 05:37:00 GMT"
}
] | 2025-06-12T00:00:00 | [
[
"Sung",
"Hakyung",
""
],
[
"Csuros",
"Karla",
""
],
[
"Sung",
"Min-Chang",
""
]
] |
2506.09022 | Daniel Shao | Daniel Shao, Richard J. Chen, Andrew H. Song, Joel Runevic, Ming Y. Lu, Tong Ding, Faisal Mahmood | Do Multiple Instance Learning Models Transfer? | ICML 2025 (Spotlight). 20 pages, 8 figures | null | null | null | cs.CV | http://creativecommons.org/licenses/by-nc-nd/4.0/ | Multiple Instance Learning (MIL) is a cornerstone approach in computational pathology (CPath) for generating clinically meaningful slide-level embeddings from gigapixel tissue images. However, MIL often struggles with small, weakly supervised clinical datasets. In contrast to fields such as NLP and conventional computer vision, where transfer learning is widely used to address data scarcity, the transferability of MIL models remains poorly understood. In this study, we systematically evaluate the transfer learning capabilities of pretrained MIL models by assessing 11 models across 21 pretraining tasks for morphological and molecular subtype prediction. Our results show that pretrained MIL models, even when trained on different organs than the target task, consistently outperform models trained from scratch. Moreover, pretraining on pancancer datasets enables strong generalization across organs and tasks, outperforming slide foundation models while using substantially less pretraining data. These findings highlight the robust adaptability of MIL models and demonstrate the benefits of leveraging transfer learning to boost performance in CPath. Lastly, we provide a resource which standardizes the implementation of MIL models and collection of pretrained model weights on popular CPath tasks, available at https://github.com/mahmoodlab/MIL-Lab | [
{
"version": "v1",
"created": "Tue, 10 Jun 2025 17:50:04 GMT"
},
{
"version": "v2",
"created": "Wed, 11 Jun 2025 11:53:38 GMT"
}
] | 2025-06-12T00:00:00 | [
[
"Shao",
"Daniel",
""
],
[
"Chen",
"Richard J.",
""
],
[
"Song",
"Andrew H.",
""
],
[
"Runevic",
"Joel",
""
],
[
"Lu",
"Ming Y.",
""
],
[
"Ding",
"Tong",
""
],
[
"Mahmood",
"Faisal",
""
]
] |
2506.09023 | Julia Guerrero-Viu | Julia Guerrero-Viu, Michael Fischer, Iliyan Georgiev, Elena Garces, Diego Gutierrez, Belen Masia and Valentin Deschaintre | Fine-Grained Spatially Varying Material Selection in Images | null | null | null | null | cs.GR cs.CV | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Selection is the first step in many image editing processes, enabling faster and simpler modifications of all pixels sharing a common modality. In this work, we present a method for material selection in images, robust to lighting and reflectance variations, which can be used for downstream editing tasks. We rely on vision transformer (ViT) models and leverage their features for selection, proposing a multi-resolution processing strategy that yields finer and more stable selection results than prior methods. Furthermore, we enable selection at two levels: texture and subtexture, leveraging a new two-level material selection (DuMaS) dataset which includes dense annotations for over 800,000 synthetic images, both on the texture and subtexture levels. | [
{
"version": "v1",
"created": "Tue, 10 Jun 2025 17:50:05 GMT"
},
{
"version": "v2",
"created": "Wed, 11 Jun 2025 17:49:53 GMT"
}
] | 2025-06-12T00:00:00 | [
[
"Guerrero-Viu",
"Julia",
""
],
[
"Fischer",
"Michael",
""
],
[
"Georgiev",
"Iliyan",
""
],
[
"Garces",
"Elena",
""
],
[
"Gutierrez",
"Diego",
""
],
[
"Masia",
"Belen",
""
],
[
"Deschaintre",
"Valentin",
""
]
] |
2506.09047 | Yaniv Nikankin | Yaniv Nikankin, Dana Arad, Yossi Gandelsman, Yonatan Belinkov | Same Task, Different Circuits: Disentangling Modality-Specific Mechanisms in VLMs | null | null | null | null | cs.CL | http://creativecommons.org/licenses/by/4.0/ | Vision-Language models (VLMs) show impressive abilities to answer questions on visual inputs (e.g., counting objects in an image), yet demonstrate higher accuracies when performing an analogous task on text (e.g., counting words in a text). We investigate this accuracy gap by identifying and comparing the \textit{circuits} - the task-specific computational sub-graphs - in different modalities. We show that while circuits are largely disjoint between modalities, they implement relatively similar functionalities: the differences lie primarily in processing modality-specific data positions (an image or a text sequence). Zooming in on the image data representations, we observe they become aligned with the higher-performing analogous textual representations only towards later layers, too late in processing to effectively influence subsequent positions. To overcome this, we patch the representations of visual data tokens from later layers back into earlier layers. In experiments with multiple tasks and models, this simple intervention closes a third of the performance gap between the modalities, on average. Our analysis sheds light on the multi-modal performance gap in VLMs and suggests a training-free approach for reducing it. | [
{
"version": "v1",
"created": "Tue, 10 Jun 2025 17:59:21 GMT"
},
{
"version": "v2",
"created": "Wed, 11 Jun 2025 11:56:44 GMT"
}
] | 2025-06-12T00:00:00 | [
[
"Nikankin",
"Yaniv",
""
],
[
"Arad",
"Dana",
""
],
[
"Gandelsman",
"Yossi",
""
],
[
"Belinkov",
"Yonatan",
""
]
] |
2506.09052 | Delower Hossain | Delower Hossain, Ehsan Saghapour, Kevin Song, Jake Y. Chen | Llama-Affinity: A Predictive Antibody Antigen Binding Model Integrating Antibody Sequences with Llama3 Backbone Architecture | 7 Pages | null | null | null | cs.LG cs.AI q-bio.QM | http://creativecommons.org/licenses/by-nc-sa/4.0/ | Antibody-facilitated immune responses are central to the body's defense against pathogens, viruses, and other foreign invaders. The ability of antibodies to specifically bind and neutralize antigens is vital for maintaining immunity. Over the past few decades, bioengineering advancements have significantly accelerated therapeutic antibody development. These antibody-derived drugs have shown remarkable efficacy, particularly in treating cancer, SARS-CoV-2, autoimmune disorders, and infectious diseases. Traditionally, experimental methods for affinity measurement have been time-consuming and expensive. With the advent of artificial intelligence, in silico medicine has been revolutionized; recent developments in machine learning, particularly the use of large language models (LLMs) for representing antibodies, have opened up new avenues for AI-based design and improved affinity prediction. Herein, we present an advanced antibody-antigen binding affinity prediction model (LlamaAffinity), leveraging an open-source Llama 3 backbone and antibody sequence data sourced from the Observed Antibody Space (OAS) database. The proposed approach shows significant improvement over existing state-of-the-art (SOTA) methods (AntiFormer, AntiBERTa, AntiBERTy) across multiple evaluation metrics. Specifically, the model achieved an accuracy of 0.9640, an F1-score of 0.9643, a precision of 0.9702, a recall of 0.9586, and an AUC-ROC of 0.9936. Moreover, this strategy unveiled higher computational efficiency, with a five-fold average cumulative training time of only 0.46 hours, significantly lower than in previous studies. | [
{
"version": "v1",
"created": "Sat, 17 May 2025 20:10:54 GMT"
}
] | 2025-06-12T00:00:00 | [
[
"Hossain",
"Delower",
""
],
[
"Saghapour",
"Ehsan",
""
],
[
"Song",
"Kevin",
""
],
[
"Chen",
"Jake Y.",
""
]
] |
2506.09053 | Sebastiano Stramaglia | Marlis Ontivero-Ortega, Gorana Mijatovic, Luca Faes, Daniele Marinazzo, and Sebastiano Stramaglia | Localizing synergies of hidden factors across complex systems: resting brain networks and HeLa gene expression profile as case studies | 11 pages, 3 figures | null | null | null | q-bio.QM physics.data-an | http://creativecommons.org/licenses/by/4.0/ | Factor analysis is a well-known statistical method to describe the variability of observed variables in terms of a smaller number of unobserved latent variables called factors. Even though latent factors are conceptually independent of each other, their influence on the observed variables is often joint and synergistic. We propose to quantify the synergy of the joint influence of factors on the observed variables using the O-information, a recently introduced metrics to assess high order dependencies in complex systems, in a new framework where latent factors and observed variables are jointly analyzed in terms of their joint informational character. Two case studies are reported: analyzing resting fMRI data, we find that DMN and FP networks show the highest synergy, consistently with their crucial role in higher cognitive functions; concerning HeLa cells, we find that the most synergistic gene is STK-12 (AURKB), suggesting that this gene is involved in controlling the HeLa cell cycle. We believe that this approach, representing a bridge between factor analysis and the field of high-order interactions, will find wide application across several domains. | [
{
"version": "v1",
"created": "Tue, 27 May 2025 20:27:35 GMT"
}
] | 2025-06-12T00:00:00 | [
[
"Ontivero-Ortega",
"Marlis",
""
],
[
"Mijatovic",
"Gorana",
""
],
[
"Faes",
"Luca",
""
],
[
"Marinazzo",
"Daniele",
""
],
[
"Stramaglia",
"Sebastiano",
""
]
] |
2506.09054 | Lachlan McGinness | Mohammad Attar, Andrew Carse, Yeming Chen, Thomas Green, Jeong-Yeon Ha, Yanbai Jin, Amy McWilliams, Theirry Panggabean, Zhengyu Peng, Lujin Sun, Jing Ru, Jiacheng She, Jialin Wang, Zilun Wei, Jiayuan Zhu, Lachlan McGinness | Particle Builder -- Learn about the Standard Model while playing against an AI | This demo has been accepted for presentation at the AIED 2025 Interactive Events Track | null | null | null | physics.ed-ph cs.HC | http://creativecommons.org/licenses/by/4.0/ | Particle Builder Online is a web-based education game designed for high school physics students. Students can play against an AI opponent or peers to familiarise themselves with the Standard Model of Particle Physics. The game is aimed at a high school level and tailored to the International Baccalaureate and the Australian Curriculum. Students from four schools in Canberra took pre/post-tests and a survey while completing a lesson where they played Particle Builder. Students' understanding of particle physics concepts improved significantly. Students found the game more enjoyable and effective than regular classroom lessons. | [
{
"version": "v1",
"created": "Tue, 27 May 2025 22:52:55 GMT"
}
] | 2025-06-12T00:00:00 | [
[
"Attar",
"Mohammad",
""
],
[
"Carse",
"Andrew",
""
],
[
"Chen",
"Yeming",
""
],
[
"Green",
"Thomas",
""
],
[
"Ha",
"Jeong-Yeon",
""
],
[
"Jin",
"Yanbai",
""
],
[
"McWilliams",
"Amy",
""
],
[
"Panggabean",
"Theirry",
""
],
[
"Peng",
"Zhengyu",
""
],
[
"Sun",
"Lujin",
""
],
[
"Ru",
"Jing",
""
],
[
"She",
"Jiacheng",
""
],
[
"Wang",
"Jialin",
""
],
[
"Wei",
"Zilun",
""
],
[
"Zhu",
"Jiayuan",
""
],
[
"McGinness",
"Lachlan",
""
]
] |
2506.09055 | Sumbul Ahmad Khan | Akshaya Rajesh, Sumbul Khan | Learn Like Feynman: Developing and Testing an AI-Driven Feynman Bot | 8 pages | 2024 IEEE International Conference on Teaching, Assessment and Learning for Engineering (TALE) | 10.1109/TALE62452.2024.10834370 | null | physics.ed-ph | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | The Feynman learning technique is an active learning strategy that helps learners simplify complex information through student-led teaching and discussion. In this paper, we present the development and usability testing of the Feynman Bot, which uses the Feynman technique to assist self-regulated learners who lack peer or instructor support. The Bot embodies the Feynman learning technique by encouraging learners to discuss their lecture material in a question-answer-driven discussion format. The Feynman Bot was developed using a large language model with Langchain in a Retrieval-Augmented-Generation framework to leverage the reasoning capability required to generate effective discussion-oriented questions. To test the Feynman bot, a controlled experiment was conducted over three days with fourteen participants. Formative and summative assessments were conducted, followed by a self-efficacy survey. We found that participants who used the Feynman Bot experienced higher learning gains than the Passive Learners' group. Moreover, Feynman Bot Learners' had a higher level of comfort with the subject after using the bot. We also found typing to be the preferred input modality method over speech, when interacting with the bot. The high learning gains and improved confidence with study material brought about by the Feynman Bot makes it a promising tool for self-regulated learners. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 08:22:19 GMT"
}
] | 2025-06-12T00:00:00 | [
[
"Rajesh",
"Akshaya",
""
],
[
"Khan",
"Sumbul",
""
]
] |
2506.09056 | Kiran Sharma Dr. | Kiran Sharmaa, Parul Khurana and Ziya Uddina | MetaInfoSci: An Integrated Web Tool for Scholarly Data Analysis | null | null | null | null | cs.DL physics.data-an | http://creativecommons.org/licenses/by/4.0/ | The exponential increase in academic publications has made it increasingly difficult for researchers to remain up to date and systematically synthesize knowledge scattered across vast and fragmented research domains. Literature reviews, particularly those supported by bibliometric methods, have become essential in organizing prior findings and guiding future research directions. While numerous tools exist for bibliometric analysis and network science, there is currently no single platform that integrates the full range of features from both domains. Researchers are often required to navigate multiple software environments, many of which lack customizable visualizations, cross-database integration, and AI-assisted result summarization. Addressing these limitations, this study introduces MetaInfoSci at www.metainfosci.com, a comprehensive, web-based platform designed to unify bibliometric, scientometric, and network analytical capabilities. The platform supports tailored query design, merges data from diverse sources, enables rich and adaptable visual outputs, and provides automated, AI-driven summaries of analytical results. This integrated approach aims to enhance the accessibility, efficiency, and depth of scientific literature analysis for scholars across disciplines. | [
{
"version": "v1",
"created": "Wed, 04 Jun 2025 07:21:22 GMT"
}
] | 2025-06-12T00:00:00 | [
[
"Sharmaa",
"Kiran",
""
],
[
"Khurana",
"Parul",
""
],
[
"Uddina",
"Ziya",
""
]
] |
2506.09057 | Bertrand Malas | B Malas (LHEEA) | Design, development and commissioning of a new towing tank at the University of Southampton: A decade of endeavour | in French language. Journ{\'e}es de l'Hydrodynamique, Nov 2024, Nantes, France | null | null | null | physics.ins-det | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | The University of Southampton in the United Kingdom is worldly renowned in the naval architecture domain. For several decades, the researchers in hydrodynamics have been discussing the need of a middle-sized towing tank to accommodate the teaching, research and commercial experiments that previously had to be outsourced. In the early 2010s, the renovation of the Boldrewood campus was decided and opened the way for the construction of this 138 m long, 6 m wide and 3.5 m deep facility. Commissioned in 2022, the tank is equipped with a 12 independent paddles wavemaker and an innovative carriage capable of speeds of up to 10 m/s. The design, construction and commissioning of the experimental facility were subject to many hurdles and delays that are described here. Details about the technical solutions and chosen equipment are provided. Validation experiments are also described. | [
{
"version": "v1",
"created": "Wed, 04 Jun 2025 08:37:21 GMT"
}
] | 2025-06-12T00:00:00 | [
[
"Malas",
"B",
"",
"LHEEA"
]
] |
2506.09058 | Ryotaro Suda | Ryotaro Suda, Kenshin Uriu, Kouki Yamamoto, Misaki Sasaki, Kento Sasaki, Mari Einaga, Katsuya Shimizu and Kensuke Kobayashi | GPa Pressure Imaging Using Nanodiamond Quantum Sensors | 10 pages, 8 figures, submitted to Journal of the Physical Society of Japan | null | null | null | cond-mat.mtrl-sci quant-ph | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | We demonstrate wide-field optical microscopy of the pressure distribution at approximately 20 GPa in a diamond anvil cell (DAC), using nitrogen-vacancy (NV) centers in nanodiamonds (NDs) as quantum sensors. Pressure and non-hydrostaticity maps are obtained by fitting optically detected magnetic resonance (ODMR) spectra with models incorporating hydrostatic and uniaxial stress conditions. Two methods for introducing NDs with a pressure-transmitting medium are compared, revealing that the embedding approach affects the degree of non-hydrostaticity. This ND-based technique offers a powerful imaging platform for probing pressure-induced phenomena and is extendable to other physical quantities such as magnetic fields. | [
{
"version": "v1",
"created": "Wed, 04 Jun 2025 12:00:12 GMT"
}
] | 2025-06-12T00:00:00 | [
[
"Suda",
"Ryotaro",
""
],
[
"Uriu",
"Kenshin",
""
],
[
"Yamamoto",
"Kouki",
""
],
[
"Sasaki",
"Misaki",
""
],
[
"Sasaki",
"Kento",
""
],
[
"Einaga",
"Mari",
""
],
[
"Shimizu",
"Katsuya",
""
],
[
"Kobayashi",
"Kensuke",
""
]
] |
2506.09059 | Dominik Fuchs | CRESST Collaboration: G. Angloher, S. Banik, A. Bento, A. Bertolini, R. Breier, C. Bucci, L. Burmeister, F. Casadei, E. Cipelli, J. Burkhart, L. Canonica, J. Dohm, F. Dominsky, S. Di Lorenzo, L. Einfalt, A. Erb, E. Fascione, F. v. Feilitzsch, S. Fichtinger, D. Fuchs, V.M. Ghete, P. Gorla, P.V. Guillaumon, D. Hauff, M. Je\v{s}kovsk\'y, J. Jochum, M. Kaznacheeva, H. Kluck, H. Kraus, B. von Krosigk, A. Langenk\"amper, M. Mancuso, B. Mauri, V. Mokina, C. Moore, P. Murali, M. Olmi, T. Ortmann, C. Pagliarone, L. Pattavina, F. Petricca, W. Potzel, P. Povinec, F. Pr\"obst, F. Pucci, F. Reindl, J. Rothe, K. Sch\"affner, J. Schieck, S. Sch\"onert, C. Schwertner, M. Stahlberg, L. Stodolsky, C. Strandhagen, R. Strauss, I. Usherov, D. Valdenaire, M. Zanirato, V. Zema and M. Macko and V. Palu\v{s}ov\'a | Observation of a low energy nuclear recoil peak in the neutron calibration data of an Al$_{2}$O$_{3}$ crystal in CRESST-III | 11 pages, 8 figures | null | null | null | physics.ins-det physics.data-an | http://creativecommons.org/licenses/by/4.0/ | The current generation of cryogenic solid state detectors used in direct dark matter and CE\textnu NS searches typically reach energy thresholds of $\mathcal{O}$(10)$\,$eV for nuclear recoils. For a reliable calibration in this energy regime a method has been proposed, providing mono-energetic nuclear recoils at low energies $\sim\,$100$\,$eV$\,$-$\,$1$\,$keV. In this work we report on the observation of a peak at (1113.6$^{+6.5}_{-6.5}$)$\,$eV in the data of an Al$_{2}$O$_{3}$ crystal in CRESST-III, which was irradiated with neutrons from an AmBe calibration source. We attribute this mono-energetic peak to the radiative capture of thermal neutrons on $^{27}$Al and the subsequent de-excitation via single $\gamma$-emission. We compare the measured results with the outcome of Geant4 simulations and investigate the possibility to make use of this effect for the energy calibration of Al$_{2}$O$_{3}$ detectors at low energies. We further investigate the possibility of a shift in the expected energy scale of this effect caused by the creation of defects in the target crystal. | [
{
"version": "v1",
"created": "Wed, 04 Jun 2025 21:47:11 GMT"
}
] | 2025-06-12T00:00:00 | [
[
"CRESST Collaboration",
"",
""
],
[
"Angloher",
"G.",
""
],
[
"Banik",
"S.",
""
],
[
"Bento",
"A.",
""
],
[
"Bertolini",
"A.",
""
],
[
"Breier",
"R.",
""
],
[
"Bucci",
"C.",
""
],
[
"Burmeister",
"L.",
""
],
[
"Casadei",
"F.",
""
],
[
"Cipelli",
"E.",
""
],
[
"Burkhart",
"J.",
""
],
[
"Canonica",
"L.",
""
],
[
"Dohm",
"J.",
""
],
[
"Dominsky",
"F.",
""
],
[
"Di Lorenzo",
"S.",
""
],
[
"Einfalt",
"L.",
""
],
[
"Erb",
"A.",
""
],
[
"Fascione",
"E.",
""
],
[
"Feilitzsch",
"F. v.",
""
],
[
"Fichtinger",
"S.",
""
],
[
"Fuchs",
"D.",
""
],
[
"Ghete",
"V. M.",
""
],
[
"Gorla",
"P.",
""
],
[
"Guillaumon",
"P. V.",
""
],
[
"Hauff",
"D.",
""
],
[
"Ješkovský",
"M.",
""
],
[
"Jochum",
"J.",
""
],
[
"Kaznacheeva",
"M.",
""
],
[
"Kluck",
"H.",
""
],
[
"Kraus",
"H.",
""
],
[
"von Krosigk",
"B.",
""
],
[
"Langenkämper",
"A.",
""
],
[
"Mancuso",
"M.",
""
],
[
"Mauri",
"B.",
""
],
[
"Mokina",
"V.",
""
],
[
"Moore",
"C.",
""
],
[
"Murali",
"P.",
""
],
[
"Olmi",
"M.",
""
],
[
"Ortmann",
"T.",
""
],
[
"Pagliarone",
"C.",
""
],
[
"Pattavina",
"L.",
""
],
[
"Petricca",
"F.",
""
],
[
"Potzel",
"W.",
""
],
[
"Povinec",
"P.",
""
],
[
"Pröbst",
"F.",
""
],
[
"Pucci",
"F.",
""
],
[
"Reindl",
"F.",
""
],
[
"Rothe",
"J.",
""
],
[
"Schäffner",
"K.",
""
],
[
"Schieck",
"J.",
""
],
[
"Schönert",
"S.",
""
],
[
"Schwertner",
"C.",
""
],
[
"Stahlberg",
"M.",
""
],
[
"Stodolsky",
"L.",
""
],
[
"Strandhagen",
"C.",
""
],
[
"Strauss",
"R.",
""
],
[
"Usherov",
"I.",
""
],
[
"Valdenaire",
"D.",
""
],
[
"Zanirato",
"M.",
""
],
[
"Zema",
"V.",
""
],
[
"Macko",
"M.",
""
],
[
"Palušová",
"V.",
""
]
] |
2506.09060 | Shiwei Jing | Yingying Cao, Sijia Zhou, Pingwei Sun, Jiayu Li, Shangrui Jiang, Shiwei Jing | Optimization of target film materials and protective coatings for sealed neutron generator | null | null | null | null | physics.ins-det cond-mat.mtrl-sci | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Magnesium target film has better thermal stability and neutron yield than titanium target, making it a potential neutron generator target film material. The radiation resistance of elemental magnesium targets is relatively weak, and their radiation resistance can be improved by alloying magnesium target films. The irradiation damage of pure magnesium targets and magnesium alloy target films was studied using SRIM. The results indicate that the irradiation damage of magnesium alloy target films (magnesium-niobium, magnesium-zirconium alloys) is lower than that of pure magnesium targets. In addition, under the same alloy ratio, the radiation resistance of magnesium-niobium alloy target film is better than that of magnesium-zirconium alloy. In order to further in-vestigate the performance of magnesium alloy target films, the incident ion energy, protective coatings (nickel oxide, aluminum oxide, palladium oxide), magnesium alloy target films, and alloy doping ratios (0.2, 0.4, 0.6, 0.8, 1.0) were changed. After calculating the effects of the above conditions on the neutron generator yield, sputtering yield, and considering irradiation damage, it was determined that a magnesium-zirconium alloy with a doping rate of 0.2 and a nickel oxide protective coating with a thickness of 7.5 nm are potential target film materials for the neutron generator. | [
{
"version": "v1",
"created": "Thu, 05 Jun 2025 08:21:25 GMT"
}
] | 2025-06-12T00:00:00 | [
[
"Cao",
"Yingying",
""
],
[
"Zhou",
"Sijia",
""
],
[
"Sun",
"Pingwei",
""
],
[
"Li",
"Jiayu",
""
],
[
"Jiang",
"Shangrui",
""
],
[
"Jing",
"Shiwei",
""
]
] |
2506.09061 | Md Rubel Ahmed | Alyssa Pinnock, Shakya Jayakody, Kawsher A Roxy, Md Rubel Ahmed | EdgeProfiler: A Fast Profiling Framework for Lightweight LLMs on Edge Using Analytical Model | 4 figures, 7 pages, IEEE conference template | null | null | null | cs.DC cs.AI cs.PF | http://creativecommons.org/licenses/by/4.0/ | This paper introduces EdgeProfiler, a fast profiling framework designed for evaluating lightweight Large Language Models (LLMs) on edge systems. While LLMs offer remarkable capabilities in natural language understanding and generation, their high computational, memory, and power requirements often confine them to cloud environments. EdgeProfiler addresses these challenges by providing a systematic methodology for assessing LLM performance in resource-constrained edge settings. The framework profiles compact LLMs, including TinyLLaMA, Gemma3.1B, Llama3.2-1B, and DeepSeek-r1-1.5B, using aggressive quantization techniques and strict memory constraints. Analytical modeling is used to estimate latency, FLOPs, and energy consumption. The profiling reveals that 4-bit quantization reduces model memory usage by approximately 60-70%, while maintaining accuracy within 2-5% of full-precision baselines. Inference speeds are observed to improve by 2-3x compared to FP16 baselines across various edge devices. Power modeling estimates a 35-50% reduction in energy consumption for INT4 configurations, enabling practical deployment on hardware such as Raspberry Pi 4/5 and Jetson Orin Nano Super. Our findings emphasize the importance of efficient profiling tailored to lightweight LLMs in edge environments, balancing accuracy, energy efficiency, and computational feasibility. | [
{
"version": "v1",
"created": "Fri, 06 Jun 2025 01:56:07 GMT"
}
] | 2025-06-12T00:00:00 | [
[
"Pinnock",
"Alyssa",
""
],
[
"Jayakody",
"Shakya",
""
],
[
"Roxy",
"Kawsher A",
""
],
[
"Ahmed",
"Md Rubel",
""
]
] |
2506.09062 | C. S. Unnikrishnan | C. S. Unnikrishnan | Information versus Physicality: On the Nature of the Wavefunctions of Quantum Mechanics | RevTeX, 2 figures | Academia Quantum 2, no. 2 (2025) | 10.20935/AcadQuant7696 | null | quant-ph | http://creativecommons.org/licenses/by-nc-nd/4.0/ | The physical states of matter and fields are represented in the quantum theory with complex valued wavefunctions, or more generally by quantum states in an abstract linear vector space. Determining the physical nature of wavefunctions remains an open problem that is at the very core of quantum mechanics, About a decade ago, Pusey, Barrett and Rudolf (PBR) claimed to prove an ontologically real status of wavefunctions by ruling out $\psi$-epistemic models. The result was obtained by associating wavefunctions to hypothetical distributions of notional physical states, and by examining whether some physical states were associated with more than one wavefunction, a criterion they chose for defining a wavefunction as `epistemic'. I show that the starting assumption in the PBR argument, of associating a wavefunction with a distribution of physical states, is flawed and contradictory to the linear structure of quantum mechanics coupled with its quadratic Born's rule. Since none of the axioms or calculations of observable statistical results in the standard quantum theory depends on specifying the physical nature of a $\psi$-function, the considerations in the PBR paper, involving a standard process of the preparation and projective measurements of quantum states, cannot address the ontological status of the wavefunctions in space and time. | [
{
"version": "v1",
"created": "Fri, 06 Jun 2025 08:33:17 GMT"
}
] | 2025-06-12T00:00:00 | [
[
"Unnikrishnan",
"C. S.",
""
]
] |
2506.09063 | Shayan Shekarforoush | Shayan Shekarforoush, David B. Lindell, Marcus A. Brubaker, David J. Fleet | Reconstructing Heterogeneous Biomolecules via Hierarchical Gaussian Mixtures and Part Discovery | 21 pages, 14 figures, Project Webpage: https://shekshaa.github.io/CryoSPIRE | null | null | null | q-bio.QM cs.CV cs.LG eess.IV | http://creativecommons.org/licenses/by/4.0/ | Cryo-EM is a transformational paradigm in molecular biology where computational methods are used to infer 3D molecular structure at atomic resolution from extremely noisy 2D electron microscope images. At the forefront of research is how to model the structure when the imaged particles exhibit non-rigid conformational flexibility and compositional variation where parts are sometimes missing. We introduce a novel 3D reconstruction framework with a hierarchical Gaussian mixture model, inspired in part by Gaussian Splatting for 4D scene reconstruction. In particular, the structure of the model is grounded in an initial process that infers a part-based segmentation of the particle, providing essential inductive bias in order to handle both conformational and compositional variability. The framework, called CryoSPIRE, is shown to reveal biologically meaningful structures on complex experimental datasets, and establishes a new state-of-the-art on CryoBench, a benchmark for cryo-EM heterogeneity methods. | [
{
"version": "v1",
"created": "Fri, 06 Jun 2025 22:38:54 GMT"
}
] | 2025-06-12T00:00:00 | [
[
"Shekarforoush",
"Shayan",
""
],
[
"Lindell",
"David B.",
""
],
[
"Brubaker",
"Marcus A.",
""
],
[
"Fleet",
"David J.",
""
]
] |
2506.09064 | Jeffery Martin | Jeffery W. Martin, B. Algohi, D. Anthony, L. Barr\'on-Palos, M. Bradley, A. Brossard, T. Bui, J. Chak, C. Davis, R. de Vries, K. Drury, D. Fujimoto, R. Fujitani, M. Gericke, P. Giampa, R. Golub, T. Hepworth, T. Higuchi, G. Ichikawa, S. Imajo, A. Jaison, B. Jamieson, M. Katotoka, S. Kawasaki, M. Kitaguchi, W. Klassen, E. Korkmaz, E. Korobkina, M. Lavvaf, T. Lindner, N. Lo, S. Longo, K. Madison, Y. Makida, J. Malcolm, J. Mammei, R. Mammei, C. Marshall, M. McCrea, E. Miller, M. Miller, K. Mishima, T. Mohammadi, T. Momose, T. Okamura, H.J. Ong, R. Patni, R. Picker, W.D. Ramsay, W. Rathnakela, J. Sato, W. Schreyer, T. Shima, H. Shimizu, S. Sidhu, S. Stargardter, P. Switzer, I. Tanihata, S. Vanbergen, W.T.H. vanOers, Y. Watanabe, A. Zahra, M.Z hao | Cryogenic systems for the TUCAN EDM experiment | proceedings of the XLVI Nuclear Physics Conference, Cocoyoc, Morelos, Mexico, January 6-9, 2025 | null | null | null | physics.ins-det nucl-ex | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | The TUCAN (TRIUMF UltraCold Advanced Neutron) Collaboration is completing a new ultracold neutron (UCN) source. The UCN source will deliver UCNs to a neutron electric dipole moment (EDM) experiment. The EDM experiment is projected to be capable of an uncertainty of $1\times 10^{-27}~e$cm, competitive with other planned projects, and a factor of ten more precise than the present world's best. The TUCAN source is based on a UCN production volume of superfluid helium (He-II), held at 1~K, and coupled to a proton-driven spallation target. The production rate in the source is expected to be in excess of $10^7$~UCN/s; since UCN losses can be small in superfluid helium, this should allow us to build up a large number of UCNs. The spallation-driven superfluid helium technology is the principal aspect making the TUCAN project unique. The superfluid production volume was recently cooled, for the first time, and successfully filled with superfluid helium. The design principles of the UCN source are described, along with some of the challenging cryogenic milestones that were recently passed. | [
{
"version": "v1",
"created": "Fri, 06 Jun 2025 23:22:04 GMT"
}
] | 2025-06-12T00:00:00 | [
[
"Martin",
"Jeffery W.",
""
],
[
"Algohi",
"B.",
""
],
[
"Anthony",
"D.",
""
],
[
"Barrón-Palos",
"L.",
""
],
[
"Bradley",
"M.",
""
],
[
"Brossard",
"A.",
""
],
[
"Bui",
"T.",
""
],
[
"Chak",
"J.",
""
],
[
"Davis",
"C.",
""
],
[
"de Vries",
"R.",
""
],
[
"Drury",
"K.",
""
],
[
"Fujimoto",
"D.",
""
],
[
"Fujitani",
"R.",
""
],
[
"Gericke",
"M.",
""
],
[
"Giampa",
"P.",
""
],
[
"Golub",
"R.",
""
],
[
"Hepworth",
"T.",
""
],
[
"Higuchi",
"T.",
""
],
[
"Ichikawa",
"G.",
""
],
[
"Imajo",
"S.",
""
],
[
"Jaison",
"A.",
""
],
[
"Jamieson",
"B.",
""
],
[
"Katotoka",
"M.",
""
],
[
"Kawasaki",
"S.",
""
],
[
"Kitaguchi",
"M.",
""
],
[
"Klassen",
"W.",
""
],
[
"Korkmaz",
"E.",
""
],
[
"Korobkina",
"E.",
""
],
[
"Lavvaf",
"M.",
""
],
[
"Lindner",
"T.",
""
],
[
"Lo",
"N.",
""
],
[
"Longo",
"S.",
""
],
[
"Madison",
"K.",
""
],
[
"Makida",
"Y.",
""
],
[
"Malcolm",
"J.",
""
],
[
"Mammei",
"J.",
""
],
[
"Mammei",
"R.",
""
],
[
"Marshall",
"C.",
""
],
[
"McCrea",
"M.",
""
],
[
"Miller",
"E.",
""
],
[
"Miller",
"M.",
""
],
[
"Mishima",
"K.",
""
],
[
"Mohammadi",
"T.",
""
],
[
"Momose",
"T.",
""
],
[
"Okamura",
"T.",
""
],
[
"Ong",
"H. J.",
""
],
[
"Patni",
"R.",
""
],
[
"Picker",
"R.",
""
],
[
"Ramsay",
"W. D.",
""
],
[
"Rathnakela",
"W.",
""
],
[
"Sato",
"J.",
""
],
[
"Schreyer",
"W.",
""
],
[
"Shima",
"T.",
""
],
[
"Shimizu",
"H.",
""
],
[
"Sidhu",
"S.",
""
],
[
"Stargardter",
"S.",
""
],
[
"Switzer",
"P.",
""
],
[
"Tanihata",
"I.",
""
],
[
"Vanbergen",
"S.",
""
],
[
"vanOers",
"W. T. H.",
""
],
[
"Watanabe",
"Y.",
""
],
[
"Zahra",
"A.",
""
],
[
"hao",
"M. Z",
""
]
] |
2506.09065 | Yassine Himeur | Abigail Copiaco, Christian Ritz, Yassine Himeur, Valsamma Eapen, Ammar Albanna and Wathiq Mansoor | Exploring Image Transforms derived from Eye Gaze Variables for Progressive Autism Diagnosis | 6 pages, 8 figures, and 1 table | null | null | null | eess.IV cs.AI cs.CV cs.HC cs.LG | http://creativecommons.org/licenses/by/4.0/ | The prevalence of Autism Spectrum Disorder (ASD) has surged rapidly over the past decade, posing significant challenges in communication, behavior, and focus for affected individuals. Current diagnostic techniques, though effective, are time-intensive, leading to high social and economic costs. This work introduces an AI-powered assistive technology designed to streamline ASD diagnosis and management, enhancing convenience for individuals with ASD and efficiency for caregivers and therapists. The system integrates transfer learning with image transforms derived from eye gaze variables to diagnose ASD. This facilitates and opens opportunities for in-home periodical diagnosis, reducing stress for individuals and caregivers, while also preserving user privacy through the use of image transforms. The accessibility of the proposed method also offers opportunities for improved communication between guardians and therapists, ensuring regular updates on progress and evolving support needs. Overall, the approach proposed in this work ensures timely, accessible diagnosis while protecting the subjects' privacy, improving outcomes for individuals with ASD. | [
{
"version": "v1",
"created": "Sat, 07 Jun 2025 19:00:17 GMT"
}
] | 2025-06-12T00:00:00 | [
[
"Copiaco",
"Abigail",
""
],
[
"Ritz",
"Christian",
""
],
[
"Himeur",
"Yassine",
""
],
[
"Eapen",
"Valsamma",
""
],
[
"Albanna",
"Ammar",
""
],
[
"Mansoor",
"Wathiq",
""
]
] |
2506.09066 | Yao Lu | Maoyu Wang, Yao Lu, Jiaqi Nie, Zeyu Wang, Yun Lin, Qi Xuan, Guan Gui | ReStNet: A Reusable & Stitchable Network for Dynamic Adaptation on IoT Devices | null | null | null | null | cs.CV cs.AI | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | With the rapid development of deep learning, a growing number of pre-trained models have been publicly available. However, deploying these fixed models in real-world IoT applications is challenging because different devices possess heterogeneous computational and memory resources, making it impossible to deploy a single model across all platforms. Although traditional compression methods, such as pruning, quantization, and knowledge distillation, can improve efficiency, they become inflexible once applied and cannot adapt to changing resource constraints. To address these issues, we propose ReStNet, a Reusable and Stitchable Network that dynamically constructs a hybrid network by stitching two pre-trained models together. Implementing ReStNet requires addressing several key challenges, including how to select the optimal stitching points, determine the stitching order of the two pre-trained models, and choose an effective fine-tuning strategy. To systematically address these challenges and adapt to varying resource constraints, ReStNet determines the stitching point by calculating layer-wise similarity via Centered Kernel Alignment (CKA). It then constructs the hybrid model by retaining early layers from a larger-capacity model and appending deeper layers from a smaller one. To facilitate efficient deployment, only the stitching layer is fine-tuned. This design enables rapid adaptation to changing budgets while fully leveraging available resources. Moreover, ReStNet supports both homogeneous (CNN-CNN, Transformer-Transformer) and heterogeneous (CNN-Transformer) stitching, allowing to combine different model families flexibly. Extensive experiments on multiple benchmarks demonstrate that ReStNet achieve flexible accuracy-efficiency trade-offs at runtime while significantly reducing training cost. | [
{
"version": "v1",
"created": "Sun, 08 Jun 2025 16:14:37 GMT"
}
] | 2025-06-12T00:00:00 | [
[
"Wang",
"Maoyu",
""
],
[
"Lu",
"Yao",
""
],
[
"Nie",
"Jiaqi",
""
],
[
"Wang",
"Zeyu",
""
],
[
"Lin",
"Yun",
""
],
[
"Xuan",
"Qi",
""
],
[
"Gui",
"Guan",
""
]
] |
2506.09067 | Zhiyu Xue | Zhiyu Xue, Reza Abbasi-Asl, Ramtin Pedarsani | Enhancing the Safety of Medical Vision-Language Models by Synthetic Demonstrations | null | null | null | null | cs.CV cs.AI | http://creativecommons.org/licenses/by/4.0/ | Generative medical vision-language models~(Med-VLMs) are primarily designed to generate complex textual information~(e.g., diagnostic reports) from multimodal inputs including vision modality~(e.g., medical images) and language modality~(e.g., clinical queries). However, their security vulnerabilities remain underexplored. Med-VLMs should be capable of rejecting harmful queries, such as \textit{Provide detailed instructions for using this CT scan for insurance fraud}. At the same time, addressing security concerns introduces the risk of over-defense, where safety-enhancing mechanisms may degrade general performance, causing Med-VLMs to reject benign clinical queries. In this paper, we propose a novel inference-time defense strategy to mitigate harmful queries, enabling defense against visual and textual jailbreak attacks. Using diverse medical imaging datasets collected from nine modalities, we demonstrate that our defense strategy based on synthetic clinical demonstrations enhances model safety without significantly compromising performance. Additionally, we find that increasing the demonstration budget alleviates the over-defense issue. We then introduce a mixed demonstration strategy as a trade-off solution for balancing security and performance under few-shot demonstration budget constraints. | [
{
"version": "v1",
"created": "Sun, 08 Jun 2025 16:26:51 GMT"
}
] | 2025-06-12T00:00:00 | [
[
"Xue",
"Zhiyu",
""
],
[
"Abbasi-Asl",
"Reza",
""
],
[
"Pedarsani",
"Ramtin",
""
]
] |
2506.09068 | Sriram Krishna | Sriram Krishna, Sravan Chittupalli, Sungjae Park | BG-HOP: A Bimanual Generative Hand-Object Prior | Presented at Agents in Interaction, from Humans to Robots, CVPR 2025 | null | null | null | cs.CV cs.LG cs.RO | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | In this work, we present BG-HOP, a generative prior that seeks to model bimanual hand-object interactions in 3D. We address the challenge of limited bimanual interaction data by extending existing single-hand generative priors, demonstrating preliminary results in capturing the joint distribution of hands and objects. Our experiments showcase the model's capability to generate bimanual interactions and synthesize grasps for given objects. We make code and models publicly available. | [
{
"version": "v1",
"created": "Sun, 08 Jun 2025 19:33:44 GMT"
}
] | 2025-06-12T00:00:00 | [
[
"Krishna",
"Sriram",
""
],
[
"Chittupalli",
"Sravan",
""
],
[
"Park",
"Sungjae",
""
]
] |
2506.09069 | Sahaj Raj Malla | Sahaj Raj Malla | Devanagari Digit Recognition using Quantum Machine Learning | 9 pages, 4 figures, arXiv preprint, code available upon request | null | null | null | quant-ph cs.CV cs.LG | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Handwritten digit recognition in regional scripts, such as Devanagari, is crucial for multilingual document digitization, educational tools, and the preservation of cultural heritage. The script's complex structure and limited annotated datasets pose significant challenges to conventional models. This paper introduces the first hybrid quantum-classical architecture for Devanagari handwritten digit recognition, combining a convolutional neural network (CNN) for spatial feature extraction with a 10-qubit variational quantum circuit (VQC) for quantum-enhanced classification. Trained and evaluated on the Devanagari Handwritten Character Dataset (DHCD), the proposed model achieves a state-of-the-art test accuracy for quantum implementation of 99.80% and a test loss of 0.2893, with an average per-class F1-score of 0.9980. Compared to equivalent classical CNNs, our model demonstrates superior accuracy with significantly fewer parameters and enhanced robustness. By leveraging quantum principles such as superposition and entanglement, this work establishes a novel benchmark for regional script recognition, highlighting the promise of quantum machine learning (QML) in real-world, low-resource language settings. | [
{
"version": "v1",
"created": "Sun, 08 Jun 2025 23:49:23 GMT"
}
] | 2025-06-12T00:00:00 | [
[
"Malla",
"Sahaj Raj",
""
]
] |
2506.09070 | Chenqi Zhang | Chenqi Zhang, Yu Feng, Jieru Zhao, Guangda Liu, Wenchao Ding, Chentao Wu, Minyi Guo | STREAMINGGS: Voxel-Based Streaming 3D Gaussian Splatting with Memory Optimization and Architectural Support | null | null | null | null | cs.GR cs.AI | http://creativecommons.org/licenses/by/4.0/ | 3D Gaussian Splatting (3DGS) has gained popularity for its efficiency and sparse Gaussian-based representation. However, 3DGS struggles to meet the real-time requirement of 90 frames per second (FPS) on resource-constrained mobile devices, achieving only 2 to 9 FPS.Existing accelerators focus on compute efficiency but overlook memory efficiency, leading to redundant DRAM traffic. We introduce STREAMINGGS, a fully streaming 3DGS algorithm-architecture co-design that achieves fine-grained pipelining and reduces DRAM traffic by transforming from a tile-centric rendering to a memory-centric rendering. Results show that our design achieves up to 45.7 $\times$ speedup and 62.9 $\times$ energy savings over mobile Ampere GPUs. | [
{
"version": "v1",
"created": "Mon, 09 Jun 2025 07:51:34 GMT"
}
] | 2025-06-12T00:00:00 | [
[
"Zhang",
"Chenqi",
""
],
[
"Feng",
"Yu",
""
],
[
"Zhao",
"Jieru",
""
],
[
"Liu",
"Guangda",
""
],
[
"Ding",
"Wenchao",
""
],
[
"Wu",
"Chentao",
""
],
[
"Guo",
"Minyi",
""
]
] |
2506.09071 | Jun Yin | Peilin Li, Jun Yin, Jing Zhong, Ran Luo, Pengyu Zeng, Miao Zhang | Segment Any Architectural Facades (SAAF):An automatic segmentation model for building facades, walls and windows based on multimodal semantics guidance | null | null | null | null | cs.CV cs.AI | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | In the context of the digital development of architecture, the automatic segmentation of walls and windows is a key step in improving the efficiency of building information models and computer-aided design. This study proposes an automatic segmentation model for building facade walls and windows based on multimodal semantic guidance, called Segment Any Architectural Facades (SAAF). First, SAAF has a multimodal semantic collaborative feature extraction mechanism. By combining natural language processing technology, it can fuse the semantic information in text descriptions with image features, enhancing the semantic understanding of building facade components. Second, we developed an end-to-end training framework that enables the model to autonomously learn the mapping relationship from text descriptions to image segmentation, reducing the influence of manual intervention on the segmentation results and improving the automation and robustness of the model. Finally, we conducted extensive experiments on multiple facade datasets. The segmentation results of SAAF outperformed existing methods in the mIoU metric, indicating that the SAAF model can maintain high-precision segmentation ability when faced with diverse datasets. Our model has made certain progress in improving the accuracy and generalization ability of the wall and window segmentation task. It is expected to provide a reference for the development of architectural computer vision technology and also explore new ideas and technical paths for the application of multimodal learning in the architectural field. | [
{
"version": "v1",
"created": "Mon, 09 Jun 2025 13:16:46 GMT"
}
] | 2025-06-12T00:00:00 | [
[
"Li",
"Peilin",
""
],
[
"Yin",
"Jun",
""
],
[
"Zhong",
"Jing",
""
],
[
"Luo",
"Ran",
""
],
[
"Zeng",
"Pengyu",
""
],
[
"Zhang",
"Miao",
""
]
] |
2506.09072 | Prashanth Sridhar | Prashanth Sridhar | A note concerning the ($S_3$) condition for roots in mixed characteristic | 2 pages | null | null | null | math.AC math.AG | http://creativecommons.org/licenses/by/4.0/ | The goal of this note is to record the following curious fact: let $S$ be an unramified regular local ring of mixed characteristic $p>0$. Let $L$ denote the quotient field of $S$ and $K=L(\omega)$ with $\omega^p\in L$. Let $R$ denote the integral closure of $S$ in $K$. Then $R$ is Cohen-Macaulay if and only if $R$ satisfies Serre's condition $(S_3)$, i.e., the obstruction to the Cohen-Macaulayness of $R$ is in codimension three. | [
{
"version": "v1",
"created": "Mon, 09 Jun 2025 16:53:12 GMT"
}
] | 2025-06-12T00:00:00 | [
[
"Sridhar",
"Prashanth",
""
]
] |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.