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 |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
2503.17359 | Jiwen Yu | Jiwen Yu, Yiran Qin, Haoxuan Che, Quande Liu, Xintao Wang, Pengfei Wan, Di Zhang, Xihui Liu | Position: Interactive Generative Video as Next-Generation Game Engine | null | null | null | null | cs.CV | http://creativecommons.org/licenses/by/4.0/ | Modern game development faces significant challenges in creativity and cost due to predetermined content in traditional game engines. Recent breakthroughs in video generation models, capable of synthesizing realistic and interactive virtual environments, present an opportunity to revolutionize game creation. In this position paper, we propose Interactive Generative Video (IGV) as the foundation for Generative Game Engines (GGE), enabling unlimited novel content generation in next-generation gaming. GGE leverages IGV's unique strengths in unlimited high-quality content synthesis, physics-aware world modeling, user-controlled interactivity, long-term memory capabilities, and causal reasoning. We present a comprehensive framework detailing GGE's core modules and a hierarchical maturity roadmap (L0-L4) to guide its evolution. Our work charts a new course for game development in the AI era, envisioning a future where AI-powered generative systems fundamentally reshape how games are created and experienced. | [
{
"version": "v1",
"created": "Fri, 21 Mar 2025 17:59:22 GMT"
},
{
"version": "v2",
"created": "Thu, 29 May 2025 16:42:53 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Yu",
"Jiwen",
""
],
[
"Qin",
"Yiran",
""
],
[
"Che",
"Haoxuan",
""
],
[
"Liu",
"Quande",
""
],
[
"Wang",
"Xintao",
""
],
[
"Wan",
"Pengfei",
""
],
[
"Zhang",
"Di",
""
],
[
"Liu",
"Xihui",
""
]
] |
2503.17820 | Zheng Lin | Zheng Lin, Nan Zhou, Chen-Xi Du, Deng-Ping Fan, Shi-Min Hu | RefCut: Interactive Segmentation with Reference Guidance | null | null | null | null | cs.CV | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Interactive segmentation aims to segment the specified target on the image with positive and negative clicks from users. Interactive ambiguity is a crucial issue in this field, which refers to the possibility of multiple compliant outcomes with the same clicks, such as selecting a part of an object versus the entire object, a single object versus a combination of multiple objects, and so on. The existing methods cannot provide intuitive guidance to the model, which leads to unstable output results and makes it difficult to meet the large-scale and efficient annotation requirements for specific targets in some scenarios. To bridge this gap, we introduce RefCut, a reference-based interactive segmentation framework designed to address part ambiguity and object ambiguity in segmenting specific targets. Users only need to provide a reference image and corresponding reference masks, and the model will be optimized based on them, which greatly reduces the interactive burden on users when annotating a large number of such targets. In addition, to enrich these two kinds of ambiguous data, we propose a new Target Disassembly Dataset which contains two subsets of part disassembly and object disassembly for evaluation. In the combination evaluation of multiple datasets, our RefCut achieved state-of-the-art performance. Extensive experiments and visualized results demonstrate that RefCut advances the field of intuitive and controllable interactive segmentation. Our code will be publicly available and the demo video is in https://www.lin-zheng.com/refcut. | [
{
"version": "v1",
"created": "Sat, 22 Mar 2025 17:14:20 GMT"
},
{
"version": "v2",
"created": "Thu, 29 May 2025 16:02:05 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Lin",
"Zheng",
""
],
[
"Zhou",
"Nan",
""
],
[
"Du",
"Chen-Xi",
""
],
[
"Fan",
"Deng-Ping",
""
],
[
"Hu",
"Shi-Min",
""
]
] |
2503.18085 | Rochana Chaturvedi | Rochana Chaturvedi, Peyman Baghershahi, Sourav Medya, Barbara Di Eugenio | Temporal Relation Extraction in Clinical Texts: A Span-based Graph Transformer Approach | Introducing a novel method for joint extraction of medical events and temporal relations from free-text, leveraging clinical LPLMs and Heterogeneous Graph Transformers, achieving a 5.5% improvement over the previous state-of-the-art and up to 8.9% on long-range relations | null | null | null | cs.CL cs.AI cs.LG | http://creativecommons.org/licenses/by/4.0/ | Temporal information extraction from unstructured text is essential for contextualizing events and deriving actionable insights, particularly in the medical domain. We address the task of extracting clinical events and their temporal relations using the well-studied I2B2 2012 Temporal Relations Challenge corpus. This task is inherently challenging due to complex clinical language, long documents, and sparse annotations. We introduce GRAPHTREX, a novel method integrating span-based entity-relation extraction, clinical large pre-trained language models (LPLMs), and Heterogeneous Graph Transformers (HGT) to capture local and global dependencies. Our HGT component facilitates information propagation across the document through innovative global landmarks that bridge distant entities. Our method improves the state-of-the-art with 5.5% improvement in the tempeval $F_1$ score over the previous best and up to 8.9% improvement on long-range relations, which presents a formidable challenge. We further demonstrate generalizability by establishing a strong baseline on the E3C corpus. This work not only advances temporal information extraction but also lays the groundwork for improved diagnostic and prognostic models through enhanced temporal reasoning. | [
{
"version": "v1",
"created": "Sun, 23 Mar 2025 14:34:49 GMT"
},
{
"version": "v2",
"created": "Thu, 29 May 2025 06:56:54 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Chaturvedi",
"Rochana",
""
],
[
"Baghershahi",
"Peyman",
""
],
[
"Medya",
"Sourav",
""
],
[
"Di Eugenio",
"Barbara",
""
]
] |
2503.18128 | Cyril Richard | P. Hardy, P. Rousselot, C. Richard, V. Boudon, X. Landsheere, A. Voute, L. Manceron, F. Kwabia Tchana | A cometary Fluorescence Model for the $\nu_3$ Vibrational Band of Cyanogen | null | null | null | null | astro-ph.EP | http://creativecommons.org/licenses/by/4.0/ | Cyanogen (\ce{C_2N_2}) is suspected for a long time to be present in comets and to contribute to the creation of CN radical. So far no observations with ground-based facilities managed to detect this species but the Rosetta mission, thanks to in situ observations with the ROSINA mass spectrometer detected this species in the coma of 67P/Churyumov-Gerasimenko. To investigate its presence from infrared spectra in other comets, we developed a fluorescence model for the $\nu_3$ fundamental band. From new laboratory high-resolution infrared spectra of cyanogen, we analyzed the region of the $\nu_3$ band of \ce{C_2N_2}, centered around \SI{4.63}{\micro\meter} (\SI{2158}{\per\centi\meter}). In addition to line positions and intensities, molecular parameters for the ground and excited vibrational state were obtained. These parameters allowed us to develop a fluorescence model for cyanogen. Line-by-line excitation rates of the $\nu_3$ band of cyanogen in cometary comae are presented. An upper limit of the abundance of cyanogen in a spectrum of comet C/2022 E3 (ZTF) is discussed. | [
{
"version": "v1",
"created": "Sun, 23 Mar 2025 16:17:12 GMT"
},
{
"version": "v2",
"created": "Tue, 01 Apr 2025 16:53:43 GMT"
},
{
"version": "v3",
"created": "Thu, 03 Apr 2025 08:41:13 GMT"
},
{
"version": "v4",
"created": "Wed, 28 May 2025 19:54:00 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Hardy",
"P.",
""
],
[
"Rousselot",
"P.",
""
],
[
"Richard",
"C.",
""
],
[
"Boudon",
"V.",
""
],
[
"Landsheere",
"X.",
""
],
[
"Voute",
"A.",
""
],
[
"Manceron",
"L.",
""
],
[
"Tchana",
"F. Kwabia",
""
]
] |
2503.18314 | Christoforos Spartalis | Christoforos N. Spartalis, Theodoros Semertzidis, Efstratios Gavves, Petros Daras | LoTUS: Large-Scale Machine Unlearning with a Taste of Uncertainty | Accepted as a main conference paper at CVPR 2025 (https://cvpr.thecvf.com/virtual/2025/poster/33292) | null | null | null | cs.LG cs.AI | http://creativecommons.org/licenses/by/4.0/ | We present LoTUS, a novel Machine Unlearning (MU) method that eliminates the influence of training samples from pre-trained models, avoiding retraining from scratch. LoTUS smooths the prediction probabilities of the model up to an information-theoretic bound, mitigating its over-confidence stemming from data memorization. We evaluate LoTUS on Transformer and ResNet18 models against eight baselines across five public datasets. Beyond established MU benchmarks, we evaluate unlearning on ImageNet1k, a large-scale dataset, where retraining is impractical, simulating real-world conditions. Moreover, we introduce the novel Retrain-Free Jensen-Shannon Divergence (RF-JSD) metric to enable evaluation under real-world conditions. The experimental results show that LoTUS outperforms state-of-the-art methods in terms of both efficiency and effectiveness. Code: https://github.com/cspartalis/LoTUS. | [
{
"version": "v1",
"created": "Mon, 24 Mar 2025 03:34:23 GMT"
},
{
"version": "v2",
"created": "Tue, 25 Mar 2025 06:23:57 GMT"
},
{
"version": "v3",
"created": "Wed, 07 May 2025 15:34:38 GMT"
},
{
"version": "v4",
"created": "Thu, 29 May 2025 13:37:02 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Spartalis",
"Christoforos N.",
""
],
[
"Semertzidis",
"Theodoros",
""
],
[
"Gavves",
"Efstratios",
""
],
[
"Daras",
"Petros",
""
]
] |
2503.18727 | David Chodounsky | David Chodounsk\'y, Monroe Eskew, Thilo Weinert | Colors of the Pseudotree | Accepted as Eurocomb'25 extended abstract with hard page limit; compromises in presentation were made | null | null | null | math.CO math.LO | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | We investigate big Ramsey degrees of finite substructures of the universal countable homogeneous meet-tree and its binary variant. We prove that structures containing antichains have infinite big Ramsey degrees, and the big Ramsey degree of a 2-element chain is at least 8 and 7 for the binary variant. We deduce that the generic C-relation does not have finite big Ramsey degrees. | [
{
"version": "v1",
"created": "Mon, 24 Mar 2025 14:36:21 GMT"
},
{
"version": "v2",
"created": "Thu, 29 May 2025 09:48:19 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Chodounský",
"David",
""
],
[
"Eskew",
"Monroe",
""
],
[
"Weinert",
"Thilo",
""
]
] |
2503.19486 | Zhengwentai Sun | Zhengwentai Sun, Chenghong Li, Hongjie Liao, Xihe Yang, Keru Zheng, Heyuan Li, Yihao Zhi, Shuliang Ning, Shuguang Cui, Xiaoguang Han | Exploring Disentangled and Controllable Human Image Synthesis: From End-to-End to Stage-by-Stage | The authors are performing a major restructuring of the paper's methodology and experiments. To avoid citation of an outdated or incomplete version, we prefer to withdraw this version. A revised version may be posted in the future | null | null | null | cs.CV | http://creativecommons.org/licenses/by-nc-nd/4.0/ | Achieving fine-grained controllability in human image synthesis is a long-standing challenge in computer vision. Existing methods primarily focus on either facial synthesis or near-frontal body generation, with limited ability to simultaneously control key factors such as viewpoint, pose, clothing, and identity in a disentangled manner. In this paper, we introduce a new disentangled and controllable human synthesis task, which explicitly separates and manipulates these four factors within a unified framework. We first develop an end-to-end generative model trained on MVHumanNet for factor disentanglement. However, the domain gap between MVHumanNet and in-the-wild data produces unsatisfactory results, motivating the exploration of virtual try-on (VTON) dataset as a potential solution. Through experiments, we observe that simply incorporating the VTON dataset as additional data to train the end-to-end model degrades performance, primarily due to the inconsistency in data forms between the two datasets, which disrupts the disentanglement process. To better leverage both datasets, we propose a stage-by-stage framework that decomposes human image generation into three sequential steps: clothed A-pose generation, back-view synthesis, and pose and view control. This structured pipeline enables better dataset utilization at different stages, significantly improving controllability and generalization, especially for in-the-wild scenarios. Extensive experiments demonstrate that our stage-by-stage approach outperforms end-to-end models in both visual fidelity and disentanglement quality, offering a scalable solution for real-world tasks. Additional demos are available on the project page: https://taited.github.io/discohuman-project/. | [
{
"version": "v1",
"created": "Tue, 25 Mar 2025 09:23:20 GMT"
},
{
"version": "v2",
"created": "Tue, 27 May 2025 07:27:18 GMT"
},
{
"version": "v3",
"created": "Thu, 29 May 2025 08:00:07 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Sun",
"Zhengwentai",
""
],
[
"Li",
"Chenghong",
""
],
[
"Liao",
"Hongjie",
""
],
[
"Yang",
"Xihe",
""
],
[
"Zheng",
"Keru",
""
],
[
"Li",
"Heyuan",
""
],
[
"Zhi",
"Yihao",
""
],
[
"Ning",
"Shuliang",
""
],
[
"Cui",
"Shuguang",
""
],
[
"Han",
"Xiaoguang",
""
]
] |
2503.20759 | Zhenghao Rao | Jeremy Kahn and Zhenghao Rao | Surface Subgroups for Cocompact Lattices of Isometries of $H^{2n}$ | 47 pages, 4 figures | null | null | null | math.GT | http://creativecommons.org/licenses/by-nc-nd/4.0/ | We prove the existence of surface subgroups within any cocompact lattice $\Gamma$ in $\mathrm{SO}(2n,1)$ for $n\geq2$. This result addresses the cases missing from the work of Hamenst\"adt in 2015, who constructed surface subgroups in cocompact lattices for all other rank-one semisimple Lie groups of non-compact type. | [
{
"version": "v1",
"created": "Wed, 26 Mar 2025 17:49:21 GMT"
},
{
"version": "v2",
"created": "Fri, 28 Mar 2025 15:44:57 GMT"
},
{
"version": "v3",
"created": "Wed, 28 May 2025 18:15:31 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Kahn",
"Jeremy",
""
],
[
"Rao",
"Zhenghao",
""
]
] |
2503.20892 | Indrani Pal | I. Pal, S. Marchesi, N. Torres-Alb\'a, I. Cox, M. Ajello, A. Banerjee, R. Silver, A. Pizzetti and K. Imam | X-ray polarization observations of NGC 2110 with IXPE | 8 pages, 4 figures, Published in A&A | A&A 697, A182 (2025) | 10.1051/0004-6361/202453641 | null | astro-ph.HE | http://creativecommons.org/licenses/by-sa/4.0/ | X-ray polarimetric observations from the Imaging X-ray Polarimeter Explorer (IXPE) is an excellent tool for probing the geometry and dynamics of X-ray emitting corona in active galactic nuclei (AGN). This work aims to investigate the geometry of the X-ray corona in the Seyfert 2 AGN, NGC 2110, using its first polarimetric observation with IXPE, conducted over a net exposure of 554 ks beginning on October 16, 2024. We performed a model-independent analysis of the 2-8 keV IXPE polarimetric observation to estimate the polarization properties of NGC 2110. Furthermore, we performed spectral and spectro-polarimetric analyses combining IXPE data with archival observations from NuSTAR, XMM-Newton, and Swift-XRT to derive detailed spectral and polarization properties. From the spectro-polarimetric analyses, an upper limit on the polarization degree (PD) of 7.6% (at the 99% confidence) was estimated in the 2-8 keV band. The spectro-polarimetric analysis in the 5.66-8 keV band yielded a looser upper limit of 27% at the 99 % confidence. Comparing the measured polarization properties, coronal parameters, and inclination angle of NGC 2110 with the Monte Carlo radiative transfer (MONK) simulations suggest that the current polarization measurements lack the sensitivity to place definitive constraints on the coronal geometry. The upper limits on PD, as derived from our analysis at the 99% confidence level, indicate that polarization remains undetected at a statistically significant level. Consequently, we are unable to determine whether the corona is elongated along the disk or more compact and spherical. Future observations with improved sensitivity will be crucial to breaking these degeneracies and providing deeper insight into the coronal structure of NGC 2110. | [
{
"version": "v1",
"created": "Wed, 26 Mar 2025 18:04:42 GMT"
},
{
"version": "v2",
"created": "Thu, 29 May 2025 03:11:12 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Pal",
"I.",
""
],
[
"Marchesi",
"S.",
""
],
[
"Torres-Albá",
"N.",
""
],
[
"Cox",
"I.",
""
],
[
"Ajello",
"M.",
""
],
[
"Banerjee",
"A.",
""
],
[
"Silver",
"R.",
""
],
[
"Pizzetti",
"A.",
""
],
[
"Imam",
"K.",
""
]
] |
2503.20960 | Arnav Arora | Arnav Arora, Srishti Yadav, Maria Antoniak, Serge Belongie, Isabelle Augenstein | Multi-Modal Framing Analysis of News | null | null | null | null | cs.CL cs.CY cs.LG | http://creativecommons.org/licenses/by/4.0/ | Automated frame analysis of political communication is a popular task in computational social science that is used to study how authors select aspects of a topic to frame its reception. So far, such studies have been narrow, in that they use a fixed set of pre-defined frames and focus only on the text, ignoring the visual contexts in which those texts appear. Especially for framing in the news, this leaves out valuable information about editorial choices, which include not just the written article but also accompanying photographs. To overcome such limitations, we present a method for conducting multi-modal, multi-label framing analysis at scale using large (vision-) language models. Grounding our work in framing theory, we extract latent meaning embedded in images used to convey a certain point and contrast that to the text by comparing the respective frames used. We also identify highly partisan framing of topics with issue-specific frame analysis found in prior qualitative work. We demonstrate a method for doing scalable integrative framing analysis of both text and image in news, providing a more complete picture for understanding media bias. | [
{
"version": "v1",
"created": "Wed, 26 Mar 2025 19:51:33 GMT"
},
{
"version": "v2",
"created": "Thu, 03 Apr 2025 13:31:11 GMT"
},
{
"version": "v3",
"created": "Thu, 29 May 2025 09:45:28 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Arora",
"Arnav",
""
],
[
"Yadav",
"Srishti",
""
],
[
"Antoniak",
"Maria",
""
],
[
"Belongie",
"Serge",
""
],
[
"Augenstein",
"Isabelle",
""
]
] |
2503.22154 | Jae-Young Yim | Jae-Young Yim, Dongwook Kim, Jae-Young Sim | Dataset Distillation of 3D Point Clouds via Distribution Matching | null | null | null | null | cs.CV | http://creativecommons.org/licenses/by-nc-nd/4.0/ | Large-scale datasets are usually required to train deep neural networks, but it increases the computational complexity hindering the practical applications. Recently, dataset distillation for images and texts has been attracting a lot of attention, that reduces the original dataset to a synthetic dataset to alleviate the computational burden of training while preserving essential task-relevant information. However, the dataset distillation for 3D point clouds remains largely unexplored, as the point clouds exhibit fundamentally different characteristics from that of images, making the dataset distillation more challenging. In this paper, we propose a distribution matching-based distillation framework for 3D point clouds that jointly optimizes the geometric structures as well as the orientations of the synthetic 3D objects. To address the semantic misalignment caused by unordered indexing of points, we introduce a Semantically Aligned Distribution Matching loss computed on the sorted features in each channel. Moreover, to address the rotation variation, we jointly learn the optimal rotation angles while updating the synthetic dataset to better align with the original feature distribution. Extensive experiments on widely used benchmark datasets demonstrate that the proposed method consistently outperforms existing dataset distillation methods, achieving superior accuracy and strong cross-architecture generalization. | [
{
"version": "v1",
"created": "Fri, 28 Mar 2025 05:15:22 GMT"
},
{
"version": "v2",
"created": "Thu, 29 May 2025 02:47:12 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Yim",
"Jae-Young",
""
],
[
"Kim",
"Dongwook",
""
],
[
"Sim",
"Jae-Young",
""
]
] |
2503.22353 | Yubo Li | Yubo Li, Yidi Miao, Xueying Ding, Ramayya Krishnan, Rema Padman | Firm or Fickle? Evaluating Large Language Models Consistency in Sequential Interactions | 8 pages, 5 figures | null | null | null | cs.CL cs.AI | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Large Language Models (LLMs) have shown remarkable capabilities across various tasks, but their deployment in high-stake domains requires consistent performance across multiple interaction rounds. This paper introduces a comprehensive framework for evaluating and improving LLM response consistency, making three key contributions. First, we propose a novel Position-Weighted Consistency (PWC) score that captures both the importance of early-stage stability and recovery patterns in multi-turn interactions. Second, we present a carefully curated benchmark dataset spanning diverse domains and difficulty levels, specifically designed to evaluate LLM consistency under various challenging follow-up scenarios. Third, we introduce Confidence-Aware Response Generation (CARG), a framework that significantly improves response stability by incorporating model confidence signals into the generation process. Empirical results demonstrate that CARG significantly improves response stability without sacrificing accuracy, underscoring its potential for reliable LLM deployment in critical applications. | [
{
"version": "v1",
"created": "Fri, 28 Mar 2025 11:49:56 GMT"
},
{
"version": "v2",
"created": "Thu, 29 May 2025 14:35:00 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Li",
"Yubo",
""
],
[
"Miao",
"Yidi",
""
],
[
"Ding",
"Xueying",
""
],
[
"Krishnan",
"Ramayya",
""
],
[
"Padman",
"Rema",
""
]
] |
2503.22622 | Jong Chul Ye | Jangho Park, Taesung Kwon, Jong Chul Ye | Zero4D: Training-Free 4D Video Generation From Single Video Using Off-the-Shelf Video Diffusion | project page: https://zero4dvid.github.io/ | null | null | null | cs.CV | http://creativecommons.org/licenses/by/4.0/ | Recently, multi-view or 4D video generation has emerged as a significant research topic. Nonetheless, recent approaches to 4D generation still struggle with fundamental limitations, as they primarily rely on harnessing multiple video diffusion models with additional training or compute-intensive training of a full 4D diffusion model with limited real-world 4D data and large computational costs. To address these challenges, here we propose the first training-free 4D video generation method that leverages the off-the-shelf video diffusion models to generate multi-view videos from a single input video. Our approach consists of two key steps: (1) By designating the edge frames in the spatio-temporal sampling grid as key frames, we first synthesize them using a video diffusion model, leveraging a depth-based warping technique for guidance. This approach ensures structural consistency across the generated frames, preserving spatial and temporal coherence. (2) We then interpolate the remaining frames using a video diffusion model, constructing a fully populated and temporally coherent sampling grid while preserving spatial and temporal consistency. Through this approach, we extend a single video into a multi-view video along novel camera trajectories while maintaining spatio-temporal consistency. Our method is training-free and fully utilizes an off-the-shelf video diffusion model, offering a practical and effective solution for multi-view video generation. | [
{
"version": "v1",
"created": "Fri, 28 Mar 2025 17:14:48 GMT"
},
{
"version": "v2",
"created": "Sun, 25 May 2025 14:05:08 GMT"
},
{
"version": "v3",
"created": "Thu, 29 May 2025 06:43:48 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Park",
"Jangho",
""
],
[
"Kwon",
"Taesung",
""
],
[
"Ye",
"Jong Chul",
""
]
] |
2503.22869 | Alexey Gavryushin | Alexey Gavryushin, Alexandros Delitzas, Luc Van Gool, Marc Pollefeys, Kaichun Mo, Xi Wang | SIGHT: Synthesizing Image-Text Conditioned and Geometry-Guided 3D Hand-Object Trajectories | null | null | null | null | cs.CV | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | When humans grasp an object, they naturally form trajectories in their minds to manipulate it for specific tasks. Modeling hand-object interaction priors holds significant potential to advance robotic and embodied AI systems in learning to operate effectively within the physical world. We introduce SIGHT, a novel task focused on generating realistic and physically plausible 3D hand-object interaction trajectories from a single image and a brief language-based task description. Prior work on hand-object trajectory generation typically relies on textual input that lacks explicit grounding to the target object, or assumes access to 3D object meshes, which are often considerably more difficult to obtain than 2D images. We propose SIGHT-Fusion, a novel diffusion-based image-text conditioned generative model that tackles this task by retrieving the most similar 3D object mesh from a database and enforcing geometric hand-object interaction constraints via a novel inference-time diffusion guidance. We benchmark our model on the HOI4D and H2O datasets, adapting relevant baselines for this novel task. Experiments demonstrate our superior performance in the diversity and quality of generated trajectories, as well as in hand-object interaction geometry metrics. | [
{
"version": "v1",
"created": "Fri, 28 Mar 2025 20:53:20 GMT"
},
{
"version": "v2",
"created": "Sat, 05 Apr 2025 09:35:08 GMT"
},
{
"version": "v3",
"created": "Thu, 29 May 2025 17:11:30 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Gavryushin",
"Alexey",
""
],
[
"Delitzas",
"Alexandros",
""
],
[
"Van Gool",
"Luc",
""
],
[
"Pollefeys",
"Marc",
""
],
[
"Mo",
"Kaichun",
""
],
[
"Wang",
"Xi",
""
]
] |
2503.22956 | Qiang Li | Leyun Gao, Cheng-en Liu, Qite Li, Chen Zhou, Qiang Li, Liangwen Chen, Xueheng Zhang, Yu Xu, and Zhiyu Sun | Probing and Knocking with Muons | Invited review for MPLA; 10 pages, 5 figures | Modern Physics Letters A (2025) 2530008 | 10.1142/S0217732325300083 | null | hep-ph hep-ex | http://creativecommons.org/licenses/by/4.0/ | We propose here a set of new methods involving probing and knocking with muons (PKMu). There is a wealth of rich physics to explore with GeV muon beams. Examples include but not limited to: muon scattering can occur at large angles, providing evidence of potential muon-philic dark matter or dark mediator candidates; muon-electron scattering can be used to detect new types of bosons associated with charged lepton flavor violation; precise measurements of GeV-scale muon-electron scattering can be employed to probe quantum correlations. | [
{
"version": "v1",
"created": "Sat, 29 Mar 2025 03:00:09 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Gao",
"Leyun",
""
],
[
"Liu",
"Cheng-en",
""
],
[
"Li",
"Qite",
""
],
[
"Zhou",
"Chen",
""
],
[
"Li",
"Qiang",
""
],
[
"Chen",
"Liangwen",
""
],
[
"Zhang",
"Xueheng",
""
],
[
"Xu",
"Yu",
""
],
[
"Sun",
"Zhiyu",
""
]
] |
2503.23509 | Tianming Liang | Tianming Liang, Haichao Jiang, Wei-Shi Zheng, Jian-Fang Hu | ReferDINO-Plus: 2nd Solution for 4th PVUW MeViS Challenge at CVPR 2025 | null | null | null | null | cs.CV | http://creativecommons.org/licenses/by/4.0/ | Referring Video Object Segmentation (RVOS) aims to segment target objects throughout a video based on a text description. This task has attracted increasing attention in the field of computer vision due to its promising applications in video editing and human-agent interaction. Recently, ReferDINO has demonstrated promising performance in this task by adapting object-level vision-language knowledge from pretrained foundational image models. In this report, we further enhance its capabilities by incorporating the advantages of SAM2 in mask quality and object consistency. In addition, to effectively balance performance between single-object and multi-object scenarios, we introduce a conditional mask fusion strategy that adaptively fuses the masks from ReferDINO and SAM2. Our solution, termed ReferDINO-Plus, achieves 60.43 \(\mathcal{J}\&\mathcal{F}\) on MeViS test set, securing 2nd place in the MeViS PVUW challenge at CVPR 2025. The code is available at: https://github.com/iSEE-Laboratory/ReferDINO-Plus. | [
{
"version": "v1",
"created": "Sun, 30 Mar 2025 16:43:04 GMT"
},
{
"version": "v2",
"created": "Thu, 29 May 2025 13:04:21 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Liang",
"Tianming",
""
],
[
"Jiang",
"Haichao",
""
],
[
"Zheng",
"Wei-Shi",
""
],
[
"Hu",
"Jian-Fang",
""
]
] |
2503.24120 | Jordan L Shivers | Jordan L. Shivers, Michael Nguyen, Aaron R. Dinner, Petia Vlahovska, and Suriyanarayanan Vaikuntanathan | Renormalized mechanics and stochastic thermodynamics of growing vesicles | null | null | null | null | cond-mat.soft physics.bio-ph | http://creativecommons.org/licenses/by/4.0/ | Uncovering the rules governing the nonequilibrium dynamics of the membranes that define biological cells is of central importance to understanding the physics of living systems. We theoretically and computationally investigate the behavior of flexible quasispherical vesicles that exchange membrane constituents, internal volume, and heat with an external reservoir. The excess chemical potential and osmotic pressure difference imposed by the reservoir act as generalized thermodynamic driving forces that modulate vesicle morphology. We show that the renormalization of membrane mechanical properties by nonequilibrium driving gives rise to a morphological transition between a weakly driven regime, in which growing vesicles remain quasispherical, and a strongly driven regime, in which vesicles accommodate rapid membrane uptake by developing surface wrinkles. Additionally, we propose a minimal vesicle growth-shape law, derived using insights from stochastic thermodynamics, that robustly describes vesicle growth dynamics even in strongly driven, far-from-equilibrium regimes. | [
{
"version": "v1",
"created": "Mon, 31 Mar 2025 14:08:15 GMT"
},
{
"version": "v2",
"created": "Wed, 28 May 2025 20:59:04 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Shivers",
"Jordan L.",
""
],
[
"Nguyen",
"Michael",
""
],
[
"Dinner",
"Aaron R.",
""
],
[
"Vlahovska",
"Petia",
""
],
[
"Vaikuntanathan",
"Suriyanarayanan",
""
]
] |
2503.24129 | Dominik Schnaus | Dominik Schnaus, Nikita Araslanov, Daniel Cremers | It's a (Blind) Match! Towards Vision-Language Correspondence without Parallel Data | Accepted to CVPR 2025, Project page: https://dominik-schnaus.github.io/itsamatch/ | null | null | null | cs.CV cs.LG | http://creativecommons.org/licenses/by/4.0/ | The platonic representation hypothesis suggests that vision and language embeddings become more homogeneous as model and dataset sizes increase. In particular, pairwise distances within each modality become more similar. This suggests that as foundation models mature, it may become possible to match vision and language embeddings in a fully unsupervised fashion, i.e. without parallel data. We present the first feasibility study, and investigate conformity of existing vision and language foundation models in the context of unsupervised, or "blind", matching. First, we formulate unsupervised matching as a quadratic assignment problem and introduce a novel heuristic that outperforms previous solvers. We also develop a technique to find optimal matching problems, for which a non-trivial match is very likely. Second, we conduct an extensive study deploying a range of vision and language models on four datasets. Our analysis reveals that for many problem instances, vision and language representations can be indeed matched without supervision. This finding opens up the exciting possibility of embedding semantic knowledge into other modalities virtually annotation-free. As a proof of concept, we showcase an unsupervised classifier, which achieves non-trivial classification accuracy without any image-text annotation. | [
{
"version": "v1",
"created": "Mon, 31 Mar 2025 14:14:25 GMT"
},
{
"version": "v2",
"created": "Thu, 29 May 2025 15:18:22 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Schnaus",
"Dominik",
""
],
[
"Araslanov",
"Nikita",
""
],
[
"Cremers",
"Daniel",
""
]
] |
2503.24380 | Xin-Chi Zhou | Xin-Chi Zhou, Bing-Chen Yao, Yongjian Wang, Yucheng Wang, Yudong Wei, Qi Zhou and Xiong-Jun Liu | The fundamental localization phases in quasiperiodic systems: A unified framework and exact results | 24 pages, 8 figures. The discussion has been significantly expanded, and a new appendix and an additional figure have been added | null | null | null | cond-mat.dis-nn cond-mat.mes-hall cond-mat.stat-mech quant-ph | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | The disordered quantum systems host three types of quantum states, the extended, localized, and critical, which bring up seven distinct fundamental phases in nature: three pure phases and four coexisting ones with mobility edges, yet a unified theory with full characterization and realization of all these phases has not been developed. Here we propose a complete and unified framework based on a spinful quasiperiodic (QP) system which realizes all the fundamental localization phases, with the exact and universal results being obtained for their characterization. First, we show that the pure phases are obtained when the chiral symmetry preserves in the proposed spinful QP model, giving a criterion for the emergence of the pure phases and otherwise the coexisting ones. Further, we uncover a novel universal mechanism for the critical states that their emergence is protected by the generalized incommensurate matrix element zeros in the spinful QP model, which considerably broadens the rigorous realizations of the exotic critical states. We then show the criteria of exact solvability for the present spinful QP system, with which we construct various exactly solvable models for all distinct localization phases. In particular, we propose two novel models, dubbed spin-selective QP lattice model and QP optical Raman lattice model, to achieve all basic types of mobility edges and all the seven fundamental phases of Anderson localization physics, respectively. The experimental scheme is proposed and studied in detail to realize these models with high feasibility. This study establishes a complete and profound theoretical framework which enables an in-depth exploration of the broad classes of all fundamental localization phenomena in QP systems, and offers key insights for constructing their exactly solvable models with experimental feasibility. | [
{
"version": "v1",
"created": "Mon, 31 Mar 2025 17:59:02 GMT"
},
{
"version": "v2",
"created": "Sun, 13 Apr 2025 17:47:12 GMT"
},
{
"version": "v3",
"created": "Wed, 30 Apr 2025 16:45:19 GMT"
},
{
"version": "v4",
"created": "Thu, 29 May 2025 06:06:26 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Zhou",
"Xin-Chi",
""
],
[
"Yao",
"Bing-Chen",
""
],
[
"Wang",
"Yongjian",
""
],
[
"Wang",
"Yucheng",
""
],
[
"Wei",
"Yudong",
""
],
[
"Zhou",
"Qi",
""
],
[
"Liu",
"Xiong-Jun",
""
]
] |
2504.00035 | Ziwei Zhang | Ziwei Zhang and Juan Wen and Wanli Peng and Zhengxian Wu and Yinghan Zhou and Yiming Xue | MiZero: The Shadowy Defender Against Text Style Infringements | null | null | null | null | cs.CR cs.AI | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | In-Context Learning (ICL) and efficient fine-tuning methods significantly enhanced the efficiency of applying Large Language Models (LLMs) to downstream tasks. However, they also raise concerns about the imitation and infringement of personal creative data. Current methods for data copyright protection primarily focuses on content security but lacks effectiveness in protecting the copyrights of text styles. In this paper, we introduce a novel implicit zero-watermarking scheme, namely MiZero. This scheme establishes a precise watermark domain to protect the copyrighted style, surpassing traditional watermarking methods that distort the style characteristics. Specifically, we employ LLMs to extract condensed-lists utilizing the designed instance delimitation mechanism. These lists guide MiZero in generating the watermark. Extensive experiments demonstrate that MiZero effectively verifies text style copyright ownership against AI imitation. | [
{
"version": "v1",
"created": "Sun, 30 Mar 2025 08:19:12 GMT"
},
{
"version": "v2",
"created": "Thu, 29 May 2025 11:19:38 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Zhang",
"Ziwei",
""
],
[
"Wen",
"Juan",
""
],
[
"Peng",
"Wanli",
""
],
[
"Wu",
"Zhengxian",
""
],
[
"Zhou",
"Yinghan",
""
],
[
"Xue",
"Yiming",
""
]
] |
2504.00186 | Olawale Salaudeen | Olawale Salaudeen, Nicole Chiou, Shiny Weng, Sanmi Koyejo | Are Domain Generalization Benchmarks with Accuracy on the Line Misspecified? | Fixed typo in covariance term in the main theorems | null | null | null | cs.LG cs.AI stat.ML | http://creativecommons.org/licenses/by-nc-nd/4.0/ | Spurious correlations are unstable statistical associations that hinder robust decision-making. Conventional wisdom suggests that models relying on such correlations will fail to generalize out-of-distribution (OOD), especially under strong distribution shifts. However, empirical evidence challenges this view as naive in-distribution empirical risk minimizers often achieve the best OOD accuracy across popular OOD generalization benchmarks. In light of these results, we propose a different perspective: many widely used benchmarks for evaluating robustness to spurious correlations are misspecified. Specifically, they fail to include shifts in spurious correlations that meaningfully impact OOD generalization, making them unsuitable for evaluating the benefit of removing such correlations. We establish conditions under which a distribution shift can reliably assess a model's reliance on spurious correlations. Crucially, under these conditions, we should not observe a strong positive correlation between in-distribution and OOD accuracy, often called "accuracy on the line." Yet, most state-of-the-art benchmarks exhibit this pattern, suggesting they do not effectively assess robustness. Our findings expose a key limitation in current benchmarks used to evaluate domain generalization algorithms, that is, models designed to avoid spurious correlations. We highlight the need to rethink how robustness to spurious correlations is assessed, identify well-specified benchmarks the field should prioritize, and enumerate strategies for designing future benchmarks that meaningfully reflect robustness under distribution shift. | [
{
"version": "v1",
"created": "Mon, 31 Mar 2025 19:50:04 GMT"
},
{
"version": "v2",
"created": "Wed, 28 May 2025 23:45:19 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Salaudeen",
"Olawale",
""
],
[
"Chiou",
"Nicole",
""
],
[
"Weng",
"Shiny",
""
],
[
"Koyejo",
"Sanmi",
""
]
] |
2504.00804 | Biao Wang | Biao Wang and Shaoyun Yi | The prime number theorem over integers of power-free polynomial values | 8 pages. Remark 4.4 is added to address the motivation of this article | null | null | null | math.NT math.DS | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Let $f(x)\in \mathbb{Z}[x]$ be an irreducible polynomial of degree $d\ge 1$. Let $k\ge2$ be an integer. The number of integers $n$ such that $f(n)$ is $k$-free is widely studied in the literature. In principle, one expects that $f(n)$ is $k$-free infinitely often, if $f$ has no fixed $k$-th power divisor. In 2022, Bergelson and Richter established a new dynamical generalization of the prime number theorem (PNT). Inspired by their work, one may expect that this generalization of the PNT also holds over integers of power-free polynomial values. In this note, we establish such variant of Bergelson and Richter's theorem for several polynomials studied by Estermann, Hooley, Heath-Brown, Booker and Browning. | [
{
"version": "v1",
"created": "Tue, 01 Apr 2025 14:00:14 GMT"
},
{
"version": "v2",
"created": "Thu, 29 May 2025 02:40:05 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Wang",
"Biao",
""
],
[
"Yi",
"Shaoyun",
""
]
] |
2504.01396 | Xi Li | Zheng Yang, Ruoxin Chen, Zhiyuan Yan, Ke-Yue Zhang, Xinghe Fu, Shuang Wu, Xiujun Shu, Taiping Yao, Shouhong Ding, Xi Li | All Patches Matter, More Patches Better: Enhance AI-Generated Image Detection via Panoptic Patch Learning | null | null | null | null | cs.CV | http://creativecommons.org/licenses/by/4.0/ | The exponential growth of AI-generated images (AIGIs) underscores the urgent need for robust and generalizable detection methods. In this paper, we establish two key principles for AIGI detection through systematic analysis: (1) All Patches Matter: Unlike conventional image classification where discriminative features concentrate on object-centric regions, each patch in AIGIs inherently contains synthetic artifacts due to the uniform generation process, suggesting that every patch serves as an important artifact source for detection. (2) More Patches Better: Leveraging distributed artifacts across more patches improves detection robustness by capturing complementary forensic evidence and reducing over-reliance on specific patches, thereby enhancing robustness and generalization. However, our counterfactual analysis reveals an undesirable phenomenon: naively trained detectors often exhibit a Few-Patch Bias, discriminating between real and synthetic images based on minority patches. We identify Lazy Learner as the root cause: detectors preferentially learn conspicuous artifacts in limited patches while neglecting broader artifact distributions. To address this bias, we propose the Panoptic Patch Learning (PPL) framework, involving: (1) Random Patch Replacement that randomly substitutes synthetic patches with real counterparts to compel models to identify artifacts in underutilized regions, encouraging the broader use of more patches; (2) Patch-wise Contrastive Learning that enforces consistent discriminative capability across all patches, ensuring uniform utilization of all patches. Extensive experiments across two different settings on several benchmarks verify the effectiveness of our approach. | [
{
"version": "v1",
"created": "Wed, 02 Apr 2025 06:32:09 GMT"
},
{
"version": "v2",
"created": "Thu, 29 May 2025 07:04:26 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Yang",
"Zheng",
""
],
[
"Chen",
"Ruoxin",
""
],
[
"Yan",
"Zhiyuan",
""
],
[
"Zhang",
"Ke-Yue",
""
],
[
"Fu",
"Xinghe",
""
],
[
"Wu",
"Shuang",
""
],
[
"Shu",
"Xiujun",
""
],
[
"Yao",
"Taiping",
""
],
[
"Ding",
"Shouhong",
""
],
[
"Li",
"Xi",
""
]
] |
2504.01792 | Limeng Qiao | Limeng Qiao, Yiyang Gan, Bairui Wang, Jie Qin, Shuang Xu, Siqi Yang, Lin Ma | UniViTAR: Unified Vision Transformer with Native Resolution | null | null | null | null | cs.CV | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Conventional Vision Transformer simplifies visual modeling by standardizing input resolutions, often disregarding the variability of natural visual data and compromising spatial-contextual fidelity. While preliminary explorations have superficially investigated native resolution modeling, existing approaches still lack systematic analysis from a visual representation perspective. To bridge this gap, we introduce UniViTAR, a family of homogeneous vision foundation models tailored for unified visual modality and native resolution scenario in the era of multimodal. Our framework first conducts architectural upgrades to the vanilla paradigm by integrating multiple advanced components. Building upon these improvements, a progressive training paradigm is introduced, which strategically combines two core mechanisms: (1) resolution curriculum learning, transitioning from fixed-resolution pretraining to native resolution tuning, thereby leveraging ViT's inherent adaptability to variable-length sequences, and (2) visual modality adaptation via inter-batch image-video switching, which balances computational efficiency with enhanced temporal reasoning. In parallel, a hybrid training framework further synergizes sigmoid-based contrastive loss with feature distillation from a frozen teacher model, thereby accelerating early-stage convergence. Finally, trained exclusively on public datasets, externsive experiments across multiple model scales from 0.3B to 1B demonstrate its effectiveness. | [
{
"version": "v1",
"created": "Wed, 02 Apr 2025 14:59:39 GMT"
},
{
"version": "v2",
"created": "Thu, 29 May 2025 16:44:34 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Qiao",
"Limeng",
""
],
[
"Gan",
"Yiyang",
""
],
[
"Wang",
"Bairui",
""
],
[
"Qin",
"Jie",
""
],
[
"Xu",
"Shuang",
""
],
[
"Yang",
"Siqi",
""
],
[
"Ma",
"Lin",
""
]
] |
2504.02154 | Chao Huang | Chao Huang, Susan Liang, Yunlong Tang, Jing Bi, Li Ma, Yapeng Tian, Chenliang Xu | FreSca: Scaling in Frequency Space Enhances Diffusion Models | Project page: https://wikichao.github.io/FreSca/ | null | null | null | cs.CV | http://creativecommons.org/licenses/by/4.0/ | Latent diffusion models (LDMs) have achieved remarkable success in a variety of image tasks, yet achieving fine-grained, disentangled control over global structures versus fine details remains challenging. This paper explores frequency-based control within latent diffusion models. We first systematically analyze frequency characteristics across pixel space, VAE latent space, and internal LDM representations. This reveals that the "noise difference" term, derived from classifier-free guidance at each step t, is a uniquely effective and semantically rich target for manipulation. Building on this insight, we introduce FreSca, a novel and plug-and-play framework that decomposes noise difference into low- and high-frequency components and applies independent scaling factors to them via spatial or energy-based cutoffs. Essentially, FreSca operates without any model retraining or architectural change, offering model- and task-agnostic control. We demonstrate its versatility and effectiveness in improving generation quality and structural emphasis on multiple architectures (e.g., SD3, SDXL) and across applications including image generation, editing, depth estimation, and video synthesis, thereby unlocking a new dimension of expressive control within LDMs. | [
{
"version": "v1",
"created": "Wed, 02 Apr 2025 22:03:11 GMT"
},
{
"version": "v2",
"created": "Fri, 23 May 2025 19:12:17 GMT"
},
{
"version": "v3",
"created": "Thu, 29 May 2025 04:20:04 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Huang",
"Chao",
""
],
[
"Liang",
"Susan",
""
],
[
"Tang",
"Yunlong",
""
],
[
"Bi",
"Jing",
""
],
[
"Ma",
"Li",
""
],
[
"Tian",
"Yapeng",
""
],
[
"Xu",
"Chenliang",
""
]
] |
2504.02298 | Xinyu Luo | Xinyu Luo, Kecheng Chen, Pao-Sheng Vincent Sun, Chris Xing Tian, Arindam Basu, Haoliang Li | SPACE: SPike-Aware Consistency Enhancement for Test-Time Adaptation in Spiking Neural Networks | null | null | null | null | cs.LG | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Spiking Neural Networks (SNNs), as a biologically plausible alternative to Artificial Neural Networks (ANNs), have demonstrated advantages in terms of energy efficiency, temporal processing, and biological plausibility. However, SNNs are highly sensitive to distribution shifts, which can significantly degrade their performance in real-world scenarios. Traditional test-time adaptation (TTA) methods designed for ANNs often fail to address the unique computational dynamics of SNNs, such as sparsity and temporal spiking behavior. To address these challenges, we propose SPike-Aware Consistency Enhancement (SPACE), the first source-free and single-instance TTA method specifically designed for SNNs. SPACE leverages the inherent spike dynamics of SNNs to maximize the consistency of spike-behavior-based local feature maps across augmented versions of a single test sample, enabling robust adaptation without requiring source data. We evaluate SPACE on multiple datasets. Furthermore, SPACE exhibits robust generalization across diverse network architectures, consistently enhancing the performance of SNNs on CNNs (such as VGG and ResNet), Transformer models, and ConvLSTM architectures. Experimental results show that SPACE outperforms state-of-the-art methods, highlighting its effectiveness and robustness in real-world settings. | [
{
"version": "v1",
"created": "Thu, 03 Apr 2025 06:05:05 GMT"
},
{
"version": "v2",
"created": "Thu, 29 May 2025 16:30:38 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Luo",
"Xinyu",
""
],
[
"Chen",
"Kecheng",
""
],
[
"Sun",
"Pao-Sheng Vincent",
""
],
[
"Tian",
"Chris Xing",
""
],
[
"Basu",
"Arindam",
""
],
[
"Li",
"Haoliang",
""
]
] |
2504.02611 | Ivor van der Hoog | Sarita de Berg, Ivor van der Hoog, Eva Rotenberg, Daniel Rutschmann, Sampson Wong | Instance-Optimal Imprecise Convex Hull | null | null | null | null | cs.CG | http://creativecommons.org/licenses/by/4.0/ | Imprecise measurements of a point set P = (p1, ..., pn) can be modelled by a family of regions F = (R1, ..., Rn), where each imprecise region Ri contains a unique point pi. A retrieval models an accurate measurement by replacing an imprecise region Ri with its corresponding point pi. We construct the convex hull of an imprecise point set in the plane, where regions in F may be retrieved at unit cost. The goal is to determine the cyclic ordering of the convex hull vertices of P as efficiently as possible. Here, efficiency is interpreted in two ways: (i) minimising the number of retrievals, and (ii) computing each retrieval location quickly.
Prior works focused on only one of these two aspects: either minimising retrievals or optimising algorithmic runtime. Our contribution is the first to simultaneously achieve both. Let r(F, P) denote the minimal number of retrievals required by any algorithm to determine the convex hull of P for a given instance (F, P). For a family F of n constant-complexity polygons, our main result is a reconstruction algorithm that performs O(r(F, P)) retrievals in O(r(F, P) log^3 n) time.
Compared to previous approaches that achieve optimal retrieval counts, we improve the runtime per retrieval by a exponential factor, from polynomial to polylogarithmic. Compared to near-linear time algorithms, we significantly reduce the number of retrievals used, and broaden the input families to include overlapping regions. We further extend our results to simple k-gons and to pairwise disjoint disks with radii in [1,k], where our runtime scales linearly with k. | [
{
"version": "v1",
"created": "Thu, 03 Apr 2025 14:12:36 GMT"
},
{
"version": "v2",
"created": "Thu, 29 May 2025 08:39:40 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"de Berg",
"Sarita",
""
],
[
"van der Hoog",
"Ivor",
""
],
[
"Rotenberg",
"Eva",
""
],
[
"Rutschmann",
"Daniel",
""
],
[
"Wong",
"Sampson",
""
]
] |
2504.02666 | Mei Li | Mei Li, Yuxiang Lu, Qinyan Dai, Suizhi Huang, Yue Ding, Hongtao Lu | BECAME: BayEsian Continual Learning with Adaptive Model MErging | Accepted by ICML 2025 | null | null | null | cs.LG cs.CV | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Continual Learning (CL) strives to learn incrementally across tasks while mitigating catastrophic forgetting. A key challenge in CL is balancing stability (retaining prior knowledge) and plasticity (learning new tasks). While representative gradient projection methods ensure stability, they often limit plasticity. Model merging techniques offer promising solutions, but prior methods typically rely on empirical assumptions and carefully selected hyperparameters. In this paper, we explore the potential of model merging to enhance the stability-plasticity trade-off, providing theoretical insights that underscore its benefits. Specifically, we reformulate the merging mechanism using Bayesian continual learning principles and derive a closed-form solution for the optimal merging coefficient that adapts to the diverse characteristics of tasks. To validate our approach, we introduce a two-stage framework named BECAME, which synergizes the expertise of gradient projection and adaptive merging. Extensive experiments show that our approach outperforms state-of-the-art CL methods and existing merging strategies. | [
{
"version": "v1",
"created": "Thu, 03 Apr 2025 15:07:28 GMT"
},
{
"version": "v2",
"created": "Thu, 29 May 2025 02:54:32 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Li",
"Mei",
""
],
[
"Lu",
"Yuxiang",
""
],
[
"Dai",
"Qinyan",
""
],
[
"Huang",
"Suizhi",
""
],
[
"Ding",
"Yue",
""
],
[
"Lu",
"Hongtao",
""
]
] |
2504.03553 | Ningyu Zhang | Shuofei Qiao, Zhisong Qiu, Baochang Ren, Xiaobin Wang, Xiangyuan Ru, Ningyu Zhang, Xiang Chen, Yong Jiang, Pengjun Xie, Fei Huang, Huajun Chen | Agentic Knowledgeable Self-awareness | ACL 2025 | null | null | null | cs.CL cs.AI cs.CV cs.LG cs.MA | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Large Language Models (LLMs) have achieved considerable performance across various agentic planning tasks. However, traditional agent planning approaches adopt a "flood irrigation" methodology that indiscriminately injects gold trajectories, external feedback, and domain knowledge into agent models. This practice overlooks the fundamental human cognitive principle of situational self-awareness during decision-making-the ability to dynamically assess situational demands and strategically employ resources during decision-making. We propose agentic knowledgeable self-awareness to address this gap, a novel paradigm enabling LLM-based agents to autonomously regulate knowledge utilization. Specifically, we propose KnowSelf, a data-centric approach that applies agents with knowledgeable self-awareness like humans. Concretely, we devise a heuristic situation judgement criterion to mark special tokens on the agent's self-explored trajectories for collecting training data. Through a two-stage training process, the agent model can switch between different situations by generating specific special tokens, achieving optimal planning effects with minimal costs. Our experiments demonstrate that KnowSelf can outperform various strong baselines on different tasks and models with minimal use of external knowledge. Code is available at https://github.com/zjunlp/KnowSelf. | [
{
"version": "v1",
"created": "Fri, 04 Apr 2025 16:03:38 GMT"
},
{
"version": "v2",
"created": "Thu, 29 May 2025 14:15:53 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Qiao",
"Shuofei",
""
],
[
"Qiu",
"Zhisong",
""
],
[
"Ren",
"Baochang",
""
],
[
"Wang",
"Xiaobin",
""
],
[
"Ru",
"Xiangyuan",
""
],
[
"Zhang",
"Ningyu",
""
],
[
"Chen",
"Xiang",
""
],
[
"Jiang",
"Yong",
""
],
[
"Xie",
"Pengjun",
""
],
[
"Huang",
"Fei",
""
],
[
"Chen",
"Huajun",
""
]
] |
2504.04413 | A. Abhishek | Abhishek, V. Suryanarayana Mummidi | Low scale leptogenesis and $TM_1$ mixing in neutrinophillic two Higgs doublet model($\nu$2HDM) with S4 flavor symmetry | 10 pages, 9 figures | null | null | null | hep-ph | http://creativecommons.org/licenses/by/4.0/ | We study a modified version of the Standard Model that includes a scalar doublet and two right-handed neutrinos, forming the neutrinophillic two higgs doublet ($\nu\text{2HDM}$) framework. For $v_2 << v_1$, this model operates at a TeV scale, bringing the RHNs within experimental reach. To further enhance its predictive power, we introduce an $S_4 \times Z_4$ flavor symmetry with five flavons, resulting in mass matrices that realize $\text{TM}_1$ mixing. The model effectively explains lepton masses and flavor mixing under the normal ordering of neutrino masses, predicting that the effective neutrino mass in 0$\nu\beta\beta$ decay lies between [4 - 5] meV, significantly lower than the sensitivity limits of current experiments. We also investigate how this framework could support low-scale leptogenesis as a natural way to explain the observed imbalance between matter and antimatter in the Universe. This work explores how neutrino physics, flavor symmetry, and baryon asymmetry are connected, providing a clear framework that links theoretical predictions with experimental possibilities. | [
{
"version": "v1",
"created": "Sun, 06 Apr 2025 08:41:55 GMT"
},
{
"version": "v2",
"created": "Wed, 28 May 2025 07:41:43 GMT"
},
{
"version": "v3",
"created": "Thu, 29 May 2025 13:46:42 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Abhishek",
"",
""
],
[
"Mummidi",
"V. Suryanarayana",
""
]
] |
2504.04532 | Moinak Bhattacharya | Moinak Bhattacharya, Saumya Gupta, Annie Singh, Chao Chen, Gagandeep Singh, Prateek Prasanna | BrainMRDiff: A Diffusion Model for Anatomically Consistent Brain MRI Synthesis | null | null | null | null | eess.IV cs.CV | http://creativecommons.org/licenses/by/4.0/ | Accurate brain tumor diagnosis relies on the assessment of multiple Magnetic Resonance Imaging (MRI) sequences. However, in clinical practice, the acquisition of certain sequences may be affected by factors like motion artifacts or contrast agent contraindications, leading to suboptimal outcome, such as poor image quality. This can then affect image interpretation by radiologists. Synthesizing high quality MRI sequences has thus become a critical research focus. Though recent advancements in controllable generative AI have facilitated the synthesis of diagnostic quality MRI, ensuring anatomical accuracy remains a significant challenge. Preserving critical structural relationships between different anatomical regions is essential, as even minor structural or topological inconsistencies can compromise diagnostic validity. In this work, we propose BrainMRDiff, a novel topology-preserving, anatomy-guided diffusion model for synthesizing brain MRI, leveraging brain and tumor anatomies as conditioning inputs. To achieve this, we introduce two key modules: Tumor+Structure Aggregation (TSA) and Topology-Guided Anatomy Preservation (TGAP). TSA integrates diverse anatomical structures with tumor information, forming a comprehensive conditioning mechanism for the diffusion process. TGAP enforces topological consistency during reverse denoising diffusion process; both these modules ensure that the generated image respects anatomical integrity. Experimental results demonstrate that BrainMRDiff surpasses existing baselines, achieving performance improvements of 23.33% on the BraTS-AG dataset and 33.33% on the BraTS-Met dataset. Code will be made publicly available soon. | [
{
"version": "v1",
"created": "Sun, 06 Apr 2025 16:16:50 GMT"
},
{
"version": "v2",
"created": "Thu, 29 May 2025 17:23:34 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Bhattacharya",
"Moinak",
""
],
[
"Gupta",
"Saumya",
""
],
[
"Singh",
"Annie",
""
],
[
"Chen",
"Chao",
""
],
[
"Singh",
"Gagandeep",
""
],
[
"Prasanna",
"Prateek",
""
]
] |
2504.05877 | Yulong Liu | Sihan Wang, Cheng Wang, Matthijs H. J. de Jong, Laure Mercier de L\'epinay, Jingwei Zhou, Mika A. Sillanp\"a\"a, Yulong Liu | Threshold-less and Flexibly Tunable Frequency Comb via Floquet Engineering | 9 pages, 4 figures | null | null | null | quant-ph cond-mat.mes-hall physics.optics | http://creativecommons.org/licenses/by-nc-nd/4.0/ | Frequency combs have revolutionized communication, metrology and spectroscopy. Numerous efforts have been dedicated to developing integrated combs, predominantly relying on Pockels or Kerr mechanisms. In this work, we propose and demonstrate a new type of frequency comb-Floquet cavity frequency comb-that does not rely on intrinsic material non-linearity. By periodically modulating the resonance frequency of a cavity, a Floquet cavity with multiple equally spaced frequency components is created. The pump tone interacts with the pre-modulated cavity, generating the output frequency comb. This approach offers a flexible tuning range and operates in a threshold-less manner, obviating the need to overcome nonlinear initiation threshold. We implement this on a microwave cavity optomechanical system on-chip. Compared to Kerr optomechanical combs, this approach efficiently generates comb with pump signal far from the cavity's intrinsic frequency, and the power required for detection is reduced by approximately a factor of \(10^6\), providing a promising platform for frequency comb generation. | [
{
"version": "v1",
"created": "Tue, 08 Apr 2025 10:05:20 GMT"
},
{
"version": "v2",
"created": "Thu, 29 May 2025 03:11:49 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Wang",
"Sihan",
""
],
[
"Wang",
"Cheng",
""
],
[
"de Jong",
"Matthijs H. J.",
""
],
[
"de Lépinay",
"Laure Mercier",
""
],
[
"Zhou",
"Jingwei",
""
],
[
"Sillanpää",
"Mika A.",
""
],
[
"Liu",
"Yulong",
""
]
] |
2504.06335 | Piercarlo Bonifacio | P. Bonifacio (1), E. Caffau (1), P. Fran\c{c}ois (2,3), M. Spite (1) ((1) LIRA - Observatoire de Paris - Universit\'e PSL - CNRS - Meudon site, (2) LIRA - Observatoire de Paris - Universit\'e PSL - CNRS - Paris site, (3) UPJV, Universit\'e de Picardie Jules Verne) | The most metal poor stars | Accepted for publication on The Astronomy & Astrophysics Review, typos corrected, one reference was missing from the latex source in the previous version | null | null | null | astro-ph.GA | http://creativecommons.org/licenses/by/4.0/ | The most metal-poor stars found in the Galaxy and in nearby galaxies are witnesses of the early evolution of the Universe. In a general picture in which we expect the metallicity to increase monotonically with time, as a result of the metal production in stars, we also expect the most metal-poor stars to be the most primitive objects accessible to our observations. The abundance ratios in these stars provide us important information on the first generations of stars that synthesised the nuclei that we observe in these stars. Because they are so primitive the modelling of their chemical inventory can be often satisfactorily achieved by assuming that all the metals were produced in a single supernova, or just a few. This is simpler than modelling the full chemical evolution, using different sources, that is necessary at higher metallicity. The price to pay for this relative ease of interpretation is that these stars are extremely rare and require specifically tailored observational strategies in order to assemble statistically significant samples of stars. In this review we try to summarise the main observational results that have been obtained in the last ten years. | [
{
"version": "v1",
"created": "Tue, 08 Apr 2025 18:00:00 GMT"
},
{
"version": "v2",
"created": "Thu, 29 May 2025 15:53:37 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Bonifacio",
"P.",
"",
"LIRA - Observatoire de Paris - Université PSL - CNRS - Meudon site"
],
[
"Caffau",
"E.",
"",
"LIRA - Observatoire de Paris - Université PSL - CNRS - Meudon site"
],
[
"François",
"P.",
"",
"LIRA - Observatoire de Paris - Université PSL - CNRS - Paris site",
"UPJV, Université de Picardie Jules Verne"
],
[
"Spite",
"M.",
"",
"LIRA - Observatoire de Paris - Université PSL - CNRS - Meudon site"
]
] |
2504.06560 | Lanrui Wang | Lanrui Wang, Mingyu Zheng, Hongyin Tang, Zheng Lin, Yanan Cao, Jingang Wang, Xunliang Cai, Weiping Wang | NeedleInATable: Exploring Long-Context Capability of Large Language Models towards Long-Structured Tables | Work in Progress | null | null | null | cs.CL | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Processing structured tabular data, particularly large and lengthy tables, constitutes a fundamental yet challenging task for large language models (LLMs). However, existing long-context benchmarks like Needle-in-a-Haystack primarily focus on unstructured text, neglecting the challenge of diverse structured tables. Meanwhile, previous tabular benchmarks mainly consider downstream tasks that require high-level reasoning abilities, and overlook models' underlying fine-grained perception of individual table cells, which is crucial for practical and robust LLM-based table applications. To address this gap, we introduce \textsc{NeedleInATable} (NIAT), a new long-context tabular benchmark that treats each table cell as a ``needle'' and requires models to extract the target cell based on cell locations or lookup questions. Our comprehensive evaluation of various LLMs and multimodal LLMs reveals a substantial performance gap between popular downstream tabular tasks and the simpler NIAT task, suggesting that they may rely on dataset-specific correlations or shortcuts to obtain better benchmark results but lack truly robust long-context understanding towards structured tables. Furthermore, we demonstrate that using synthesized NIAT training data can effectively improve performance on both NIAT task and downstream tabular tasks, which validates the importance of NIAT capability for LLMs' genuine table understanding ability. Our data, code and models will be released to facilitate future research. | [
{
"version": "v1",
"created": "Wed, 09 Apr 2025 03:46:56 GMT"
},
{
"version": "v2",
"created": "Thu, 29 May 2025 03:31:02 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Wang",
"Lanrui",
""
],
[
"Zheng",
"Mingyu",
""
],
[
"Tang",
"Hongyin",
""
],
[
"Lin",
"Zheng",
""
],
[
"Cao",
"Yanan",
""
],
[
"Wang",
"Jingang",
""
],
[
"Cai",
"Xunliang",
""
],
[
"Wang",
"Weiping",
""
]
] |
2504.06592 | Kazuki Watanabe | Mayuko Kori, Kazuki Watanabe | On Coalgebraic Product Constructions for Markov Chains and Automata | Preprint | null | null | null | cs.LO | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Verifying traces of systems is a central topic in formal verification. We study model checking of Markov chains (MCs) against temporal properties represented as (finite) automata. For instance, given an MC and a deterministic finite automaton (DFA), a simple but practically useful model checking problem asks for the probability of traces on the MC that are accepted by the DFA. A standard approach to solving this problem constructs a product MC of the given MC and DFA, reducing the task to a simple reachability probability problem on the resulting product MC.
In this paper, on top of our recent development of coalgebraic framework, we first present a no-go theorem for product constructions, showing a case when we cannot do product constructions for model checking. Specifically, we show that there are no coalgebraic product MCs of MCs and nondeterministic finite automata for computing the probability of the accepting traces. This no-go theorem is established via a characterisation of natural transformations between certain functors that determine the type of branching, including nondeterministic or probabilistic branching.
Second, we present a coalgebraic product construction of MCs and multiset finite automata (MFAs) as a new instance within our framework. This construction addresses a model checking problem that asks for the expected number of accepting runs on MFAs over traces of MCs. The problem is reduced to solving linear equations, which is solvable in polynomial-time under a reasonable assumption that ensures the finiteness of the solution. | [
{
"version": "v1",
"created": "Wed, 09 Apr 2025 05:41:29 GMT"
},
{
"version": "v2",
"created": "Thu, 29 May 2025 11:47:58 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Kori",
"Mayuko",
""
],
[
"Watanabe",
"Kazuki",
""
]
] |
2504.06940 | Letian Tang | Letian Tang | More-efficient Quantum Multivariate Mean Value Estimator from Generalized Grover Operator | 39 pages, 0 figures | null | null | null | quant-ph cs.CC | http://creativecommons.org/licenses/by/4.0/ | In this work, we present an efficient algorithm for multivariate mean value estimation. Our algorithm outperforms previous work by polylog factors and nearly saturates the known lower bound. More formally, given a random vector $\vec{X}$ of dimension $d$, we find an algorithm that uses $O\left(n \log \frac{d}{\delta}\right)$ samples to find a mean estimate that $\vec{\tilde{\mu}}$ that differs from the true mean $\vec{\mu}$ by $\frac{\sqrt{\text{tr } \Sigma}}{n}$ in $\ell^\infty$ norm and hence $\frac{\sqrt{d \text{ tr } \Sigma}}{n}$ in $\ell^2$ norm, where $\Sigma$ is the covariance matrix of the components of the random vector. We also presented another algorithm that uses smaller memory but costs an extra $d^\frac{1}{4}$ in complexity. Consider the Grover operator, the unitary operator used in Grover's algorithm. It contains an oracle that uses a $\pm 1$ phase for each candidate for the search space. Previous work has demonstrated that when we substitute the oracle in Grover operator with generic phases, it ended up being a good mean value estimator in some mathematical notion. We used this idea to build our algorithm. Our result remains not exactly optimal due to a $\log \frac{d}{\delta}$ term in our complexity, as opposed to something nicer such as $\log \frac{1}{\delta}$; This comes from the phase estimation primitive in our algorithm. So far, this primitive is the only major known method to tackle the problem, and moving beyond this idea seems hard. Our results demonstrates that the methodology with generalized Grover operator can be used develop the optimal algorithm without polylog overhead for different tasks relating to mean value estimation. | [
{
"version": "v1",
"created": "Wed, 09 Apr 2025 14:48:23 GMT"
},
{
"version": "v2",
"created": "Thu, 10 Apr 2025 19:50:10 GMT"
},
{
"version": "v3",
"created": "Thu, 29 May 2025 17:01:27 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Tang",
"Letian",
""
]
] |
2504.07074 | Alberto Saldana-Lopez | A. Saldana-Lopez (SU), M. J. Hayes (SU), A. Le Reste (UMN), C. Scarlata (UMN), J. Melinder (SU), A. Henry (STScI), F. Leclercq (CNRS), T. Garel (UniGE), R. Amorin, H. Atek, O. Bait, C. A. Carr, J. Chisholm, S. R. Flury, T. M. Heckman, A. E. Jaskot, I. Jung, Z. Ji, L. Komarova, Y-H. Lin, M. S. Oey, G. Ostlin, L. Pentericci, A. Runnholm, D. Schaerer, T. X. Thuan, X. Xu | The Lyman-alpha and Continuum Origins Survey II: the connection between the escape of ionizing radiation and Lyman-alpha halos in star-forming galaxies | Submitted to ApJ, 18 pages, 12 figures, 1 table | null | null | null | astro-ph.GA | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | One of the current challenges in galaxy evolution studies is to establish the mechanisms that govern the escape of ionizing radiation from galaxies. In this work, we investigate the connection between Lyman Continuum (LyC) escape and the conditions of the Circumgalactic Medium (CGM), as probed by Ly$\alpha$ halos (LAHs) in emission. We use Ly$\alpha$ and UV continuum imaging data from the Lyman alpha and Continuum Origins Survey (LaCOS), targeting 42 nearby ($z \simeq 0.3$), star-forming galaxies with LyC observations (escape fractions of $f_{\rm esc}^{\rm LyC} \simeq 0.01-0.49$). LaCOS galaxies show extended Ly$\alpha$ emission ubiquitously, with LyC emitters (LCEs) having more compact Ly$\alpha$ morphologies than non-LCEs, and Ly$\alpha$ spatial offsets that do not exceed the extent of the UV continuum. We model the diffuse LAHs using a combined S\'ersic plus exponential 2D profile, and find that the characteristic scale length of the Ly$\alpha$ halo is ten times larger than the UV, on average. We unveil a significant anti-correlation between $f_{\rm esc}^{\rm LyC}$ and the Ly$\alpha$ Halo Fraction (HF, or contribution of the halo to the total Ly$\alpha$ luminosity), that we propose as a new LyC indicator. Our observations show that halo scale lengths and HFs both scale positively with the optical depth of the neutral gas in the ISM, revealing a picture in which Ly$\alpha$ and LyC photons in LCEs either emerge directly from the central starbursts or escape isotropically and, in the case of Ly$\alpha$, minimize the number of scattering interactions in a less-extended CGM. The properties of LAHs in LaCOS resemble those of LAHs at $z \geq 3$, suggesting a lack of evolution in the $f_{\rm esc}^{\rm LyC}$ predictors that rely on the spatial properties of Ly$\alpha$, and ensuring the applicability of these indicators to observations of high-redshift galaxies. | [
{
"version": "v1",
"created": "Wed, 09 Apr 2025 17:46:13 GMT"
},
{
"version": "v2",
"created": "Wed, 28 May 2025 19:05:01 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Saldana-Lopez",
"A.",
"",
"SU"
],
[
"Hayes",
"M. J.",
"",
"SU"
],
[
"Reste",
"A. Le",
"",
"UMN"
],
[
"Scarlata",
"C.",
"",
"UMN"
],
[
"Melinder",
"J.",
"",
"SU"
],
[
"Henry",
"A.",
"",
"STScI"
],
[
"Leclercq",
"F.",
"",
"CNRS"
],
[
"Garel",
"T.",
"",
"UniGE"
],
[
"Amorin",
"R.",
""
],
[
"Atek",
"H.",
""
],
[
"Bait",
"O.",
""
],
[
"Carr",
"C. A.",
""
],
[
"Chisholm",
"J.",
""
],
[
"Flury",
"S. R.",
""
],
[
"Heckman",
"T. M.",
""
],
[
"Jaskot",
"A. E.",
""
],
[
"Jung",
"I.",
""
],
[
"Ji",
"Z.",
""
],
[
"Komarova",
"L.",
""
],
[
"Lin",
"Y-H.",
""
],
[
"Oey",
"M. S.",
""
],
[
"Ostlin",
"G.",
""
],
[
"Pentericci",
"L.",
""
],
[
"Runnholm",
"A.",
""
],
[
"Schaerer",
"D.",
""
],
[
"Thuan",
"T. X.",
""
],
[
"Xu",
"X.",
""
]
] |
2504.07472 | Jinlong He | Jinlong He, Binru Huang, Changwei Xia, Hengqin Yang, Jiwei Yan, Jun Yan | HACMony: Automatically Detecting Hopping-related Audio-stream Conflict Issues on HarmonyOS | null | null | null | null | cs.SE | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | HarmonyOS is emerging as a popular distributed operating system for diverse mobile devices. One of its standout features is app-hopping, which allows users to seamlessly transition apps across different HarmonyOS devices. However, when apps playing audio streams hop between devices, they can easily trigger Hopping-related Audio-stream Conflict (HAC) scenarios. Improper resolution of HAC will lead to significant HAC issues, which are harder to detect compared to single-device audio-stream conflicts, due to the unclear semantics of HarmonyOS's app-hopping mechanism and the lack of effective multi-app hopping testing methods. To fill the gap, this paper introduces an automated and efficient approach to detecting HAC issues. We formalized the operational semantics of HarmonyOS's app-hopping mechanism for audio streams for the first time. Leveraging this formalization, we designed an Audio Service Transition Graph (ASTG) to model the behaviors of audio-API-related services and proposed a model-based approach to detect HAC issues automatically. Our techniques were implemented in a tool, HACMony, and evaluated on 20 real-world HarmonyOS apps. Experimental results reveal that 11 of the 20 apps exhibit HAC issues. Additionally, we summarized the detected issues into two typical types, namely MoD and MoR, and analyzed their characteristics to assist and guide both app and OS developers. | [
{
"version": "v1",
"created": "Thu, 10 Apr 2025 05:55:08 GMT"
},
{
"version": "v2",
"created": "Thu, 29 May 2025 03:22:21 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"He",
"Jinlong",
""
],
[
"Huang",
"Binru",
""
],
[
"Xia",
"Changwei",
""
],
[
"Yang",
"Hengqin",
""
],
[
"Yan",
"Jiwei",
""
],
[
"Yan",
"Jun",
""
]
] |
2504.07873 | O.A. Veliev | O. A. Veliev | On the Spectrality of the Differential Operators with Periodic Coefficients | null | null | null | null | math.SP | http://creativecommons.org/licenses/by/4.0/ | In this paper, we establish a condition on the coefficients of differential operators generated in the space of square-integrable functions on the entire real line by an ordinary differential expression with periodic, complex-valued coefficients, under which the operator is a spectral operator in the sense of Dunford [1]. | [
{
"version": "v1",
"created": "Thu, 10 Apr 2025 15:46:53 GMT"
},
{
"version": "v2",
"created": "Wed, 16 Apr 2025 10:01:09 GMT"
},
{
"version": "v3",
"created": "Wed, 21 May 2025 17:02:50 GMT"
},
{
"version": "v4",
"created": "Thu, 29 May 2025 15:11:29 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Veliev",
"O. A.",
""
]
] |
2504.07934 | Xiyao Wang | Xiyao Wang, Zhengyuan Yang, Chao Feng, Hongjin Lu, Linjie Li, Chung-Ching Lin, Kevin Lin, Furong Huang, Lijuan Wang | SoTA with Less: MCTS-Guided Sample Selection for Data-Efficient Visual Reasoning Self-Improvement | 27 pages, 5 figures | null | null | null | cs.CV | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | We introduce ThinkLite-VL, a family of visual reasoning models that achieve state-of-the-art (SoTA) performance using an order of magnitude fewer training samples, relying purely on reinforcement fine-tuning (RFT) self-improvement without any knowledge distillation. Our central insight is that sample difficulty critically influences RFT effectiveness: appropriately challenging examples can drive substantial reasoning improvements, even in low-data regimes. However, quantifying sample difficulty in a reliable and scalable manner remains non-trivial. To address this, we repurpose Monte Carlo Tree Search (MCTS) to measure sample difficulty via the number of reasoning iterations a vision-language model (VLM) requires to solve each instance. This MCTS-based selection procedure identifies samples that induce deeper reasoning while remaining solvable, allowing us to filter a high-quality subset from 70k open-source examples spanning math, natural image understanding, and chart comprehension. Using this approach, we select just 11k challenging samples for RFT on Qwen2.5-VL-7B-Instruct and 7.5k samples for Qwen2.5-VL-72B-Instruct. The resulting models, ThinkLite-VL-7B and ThinkLite-VL-72B, significantly outperform their respective base models across eight visual reasoning benchmarks. In particular, ThinkLite-VL-7B improves the average performance of Qwen2.5-VL-7B-Instruct by 7\% and surpasses all existing 7B-level models, as well as much larger models such as GPT-4o, O1 and Qwen2.5-VL-72B, achieving a new SoTA score of 75.1 on MathVista. ThinkLite-VL-72B further advances the SoTA frontier, achieving an accuracy of 79.7 on MathVista and an average benchmark improvement of 4.42 over the open-source SOTA. These results demonstrate that MCTS-guided difficulty filtering provides a scalable and effective path toward data-efficient self-improvement in multimodal reasoning. | [
{
"version": "v1",
"created": "Thu, 10 Apr 2025 17:49:05 GMT"
},
{
"version": "v2",
"created": "Wed, 28 May 2025 21:55:27 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Wang",
"Xiyao",
""
],
[
"Yang",
"Zhengyuan",
""
],
[
"Feng",
"Chao",
""
],
[
"Lu",
"Hongjin",
""
],
[
"Li",
"Linjie",
""
],
[
"Lin",
"Chung-Ching",
""
],
[
"Lin",
"Kevin",
""
],
[
"Huang",
"Furong",
""
],
[
"Wang",
"Lijuan",
""
]
] |
2504.08120 | Daniil Larionov | Daniil Larionov, Sotaro Takeshita, Ran Zhang, Yanran Chen, Christoph Leiter, Zhipin Wang, Christian Greisinger, Steffen Eger | DeepSeek vs. o3-mini: How Well can Reasoning LLMs Evaluate MT and Summarization? | null | null | null | null | cs.CL | http://creativecommons.org/licenses/by-sa/4.0/ | Reasoning-enabled large language models (LLMs) excel in logical tasks, yet their utility for evaluating natural language generation remains unexplored. This study systematically compares reasoning LLMs with non-reasoning counterparts across machine translation and text summarization evaluation tasks. We evaluate eight models spanning state-of-the-art reasoning models (DeepSeek-R1, OpenAI o3), their distilled variants (8B-70B parameters), and equivalent non-reasoning LLMs. Experiments on WMT23 and SummEval benchmarks reveal architecture and task-dependent benefits: OpenAI o3-mini models show improved performance with increased reasoning on MT, while DeepSeek-R1 and generally underperforms compared to its non-reasoning variant except in summarization consistency evaluation. Correlation analysis demonstrates that reasoning token usage correlates with evaluation quality only in specific models, while almost all models generally allocate more reasoning tokens when identifying more quality issues. Distillation maintains reasonable performance up to 32B parameter models but degrades substantially at 8B scale. This work provides the first assessment of reasoning LLMs for NLG evaluation and comparison to non-reasoning models. We share our code to facilitate further research: https://github.com/NL2G/reasoning-eval. | [
{
"version": "v1",
"created": "Thu, 10 Apr 2025 20:39:18 GMT"
},
{
"version": "v2",
"created": "Thu, 29 May 2025 13:35:47 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Larionov",
"Daniil",
""
],
[
"Takeshita",
"Sotaro",
""
],
[
"Zhang",
"Ran",
""
],
[
"Chen",
"Yanran",
""
],
[
"Leiter",
"Christoph",
""
],
[
"Wang",
"Zhipin",
""
],
[
"Greisinger",
"Christian",
""
],
[
"Eger",
"Steffen",
""
]
] |
2504.08241 | S. Venkatesan Diwakar | Saumyakanta Mishra and S. V. Diwakar | Onset of thermo-convective instabilities in two-layer binary fluid systems | null | null | null | null | physics.flu-dyn | http://creativecommons.org/licenses/by/4.0/ | The current work analyses the onset characteristics of buoyancy and thermocapillary-driven instabilities in two-layer binary fluid systems near their upper critical solution temperature (UCST). The dynamics of the binary fluids are modelled here via a diffuse interface approach (phase-field method) involving a modified free energy formulation to capture the temperature-dependent solubility and interfacial width. Using spectral collocation-based discretization and a suitable grid mapping strategy, the present work accurately predicts the neutral curves for different fluid combinations that adhere to the concept of balanced contrasts. In the case of pure buoyancy-driven (Rayleigh-Benard) convection, the parametric range for oscillatory onset is found to shrink when the system approaches USCT, as the increased solubility results in less favourable conditions for oscillatory onset. The marginal stability curves of each fluid combination exhibit their own drift pattern based on the thermo-physical and transport properties. For systems with added thermocapillarity effects (Rayleigh-Benard-Marangoni convection), the changing solubilities and the interfacial thickness act along with the interfacial tension to exhibit a dual role that results in system-specific expansion/shrinkage of the parametric space for oscillatory flow onset. | [
{
"version": "v1",
"created": "Fri, 11 Apr 2025 03:55:36 GMT"
},
{
"version": "v2",
"created": "Thu, 29 May 2025 10:01:50 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Mishra",
"Saumyakanta",
""
],
[
"Diwakar",
"S. V.",
""
]
] |
2504.08297 | Rohan Singh | Bhaskar De and Pradeep Kumar and Krishna K. Maurya and Rishabh Tripathi and Rohan Singh | Quantitative Lineshape Analysis for Arbitrary Inhomogeneity in Two-Dimensional Coherent Spectroscopy | null | null | null | null | physics.optics cond-mat.mtrl-sci | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Two-dimensional coherent spectroscopy (2DCS) provides simultaneous measurement of homogeneous and inhomogeneous linewidths through quantitative lineshape analysis. However, conventional lineshape analysis methods assume Gaussian inhomogeneity, limiting its applicability to systems with non-Gaussian inhomogeneity. We present a quantitative lineshape analysis method incorporating arbitrary inhomogeneity using a bivariate spectral distribution function in 2DCS simulations. An algorithm is developed to extract the homogeneous linewidth and arbitrary inhomogeneous distribution from experimentally-measured 2D spectrum. We demonstrate this framework for a quantum-well-exciton resonance with non-Gaussian inhomogeneity. This work broadens the scope of quantitative lineshape analysis for studying materials with non-Gaussian inhomogeneity. | [
{
"version": "v1",
"created": "Fri, 11 Apr 2025 06:55:53 GMT"
},
{
"version": "v2",
"created": "Thu, 29 May 2025 12:21:02 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"De",
"Bhaskar",
""
],
[
"Kumar",
"Pradeep",
""
],
[
"Maurya",
"Krishna K.",
""
],
[
"Tripathi",
"Rishabh",
""
],
[
"Singh",
"Rohan",
""
]
] |
2504.09383 | Azur {\DJ}onlagi\'c | Azur {\DJ}onlagi\'c | Numerical Calculation of Periods on Schoen's Class of Calabi-Yau Threefolds | 39 pages, 5 figures | null | null | null | math.AG | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Through classical modularity conjectures, the period integrals of a holomorphic $3$-form on a rigid Calabi-Yau threefold are interesting from the perspective of number theory. Although the (approximate) values of these integrals would be very useful for studying such relations, they are difficult to calculate and generally not known outside of the rare cases in which we can express them exactly.
In this paper, we present an efficient numerical method to compute such periods on a wide class of Calabi-Yau threefolds constructed by small resolutions of fiber products of elliptic surfaces over $\mathbf{P}^1$, introduced by C. Schoen in his 1988 paper. Many example results are given, which can easily be calculated with arbitrary precision. We provide tables in which each result is written with precision of 30 decimal places and then compared to period integrals of the appropriate modular form, to confirm accuracy. | [
{
"version": "v1",
"created": "Sun, 13 Apr 2025 00:09:54 GMT"
},
{
"version": "v2",
"created": "Thu, 29 May 2025 13:19:47 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Đonlagić",
"Azur",
""
]
] |
2504.09445 | Yu-Xiao Liu | Yu-Qi Dong, Xiao-Bin Lai, Yu-Zhi Fan, Yu-Xiao Liu | New gravitational wave polarization modes in the torsionless spacetime | 24 pages+6 appendices, 3 figures, improve the structure of the manuscript | null | null | null | gr-qc hep-th | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | In this study, we investigate the polarization properties of gravitational waves within a torsionless spacetime framework, as described by the Palatini formalism. Our analysis uncovers the presence of two novel polarization modes, referred to as shear modes, which extend beyond the traditional set of six modes in a four-dimensional Riemannian spacetime. These shear modes, uniquely driven by vector degrees of freedom associated with non-metricity, are classified as vector modes, and their detection provides a unique opportunity to explore the fundamental structure of spacetime and to test gravity theories. These modes extend the standard gravitational wave polarization paradigm and provide novel observational signatures for gravitational wave detectors. | [
{
"version": "v1",
"created": "Sun, 13 Apr 2025 06:03:01 GMT"
},
{
"version": "v2",
"created": "Thu, 29 May 2025 09:27:32 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Dong",
"Yu-Qi",
""
],
[
"Lai",
"Xiao-Bin",
""
],
[
"Fan",
"Yu-Zhi",
""
],
[
"Liu",
"Yu-Xiao",
""
]
] |
2504.09570 | Biao Fu | Biao Fu, Minpeng Liao, Kai Fan, Chengxi Li, Liang Zhang, Yidong Chen, Xiaodong Shi | LLMs Can Achieve High-quality Simultaneous Machine Translation as Efficiently as Offline | Camera ready version for ACL 2025 Findings | null | null | null | cs.CL | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | When the complete source sentence is provided, Large Language Models (LLMs) perform excellently in offline machine translation even with a simple prompt "Translate the following sentence from [src lang] into [tgt lang]:". However, in many real scenarios, the source tokens arrive in a streaming manner and simultaneous machine translation (SiMT) is required, then the efficiency and performance of decoder-only LLMs are significantly limited by their auto-regressive nature. To enable LLMs to achieve high-quality SiMT as efficiently as offline translation, we propose a novel paradigm that includes constructing supervised fine-tuning (SFT) data for SiMT, along with new training and inference strategies. To replicate the token input/output stream in SiMT, the source and target tokens are rearranged into an interleaved sequence, separated by special tokens according to varying latency requirements. This enables powerful LLMs to learn read and write operations adaptively, based on varying latency prompts, while still maintaining efficient auto-regressive decoding. Experimental results show that, even with limited SFT data, our approach achieves state-of-the-art performance across various SiMT benchmarks, and preserves the original abilities of offline translation. Moreover, our approach generalizes well to document-level SiMT setting without requiring specific fine-tuning, even beyond the offline translation model. | [
{
"version": "v1",
"created": "Sun, 13 Apr 2025 13:45:53 GMT"
},
{
"version": "v2",
"created": "Thu, 29 May 2025 13:06:43 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Fu",
"Biao",
""
],
[
"Liao",
"Minpeng",
""
],
[
"Fan",
"Kai",
""
],
[
"Li",
"Chengxi",
""
],
[
"Zhang",
"Liang",
""
],
[
"Chen",
"Yidong",
""
],
[
"Shi",
"Xiaodong",
""
]
] |
2504.09851 | Aikaterini Maria Panteleaki | Aikaterini Maria Panteleaki, Konstantinos Balaskas, Georgios Zervakis, Hussam Amrouch, Iraklis Anagnostopoulos | Carbon-Efficient 3D DNN Acceleration: Optimizing Performance and Sustainability | IEEE Computer Society Annual Symposium on VLSI (ISVLSI) 2025 | null | null | null | cs.AR cs.AI | http://creativecommons.org/licenses/by/4.0/ | As Deep Neural Networks (DNNs) continue to drive advancements in artificial intelligence, the design of hardware accelerators faces growing concerns over embodied carbon footprint due to complex fabrication processes. 3D integration improves performance but introduces sustainability challenges, making carbon-aware optimization essential. In this work, we propose a carbon-efficient design methodology for 3D DNN accelerators, leveraging approximate computing and genetic algorithm-based design space exploration to optimize Carbon Delay Product (CDP). By integrating area-efficient approximate multipliers into Multiply-Accumulate (MAC) units, our approach effectively reduces silicon area and fabrication overhead while maintaining high computational accuracy. Experimental evaluations across three technology nodes (45nm, 14nm, and 7nm) show that our method reduces embodied carbon by up to 30% with negligible accuracy drop. | [
{
"version": "v1",
"created": "Mon, 14 Apr 2025 03:48:37 GMT"
},
{
"version": "v2",
"created": "Thu, 29 May 2025 16:57:22 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Panteleaki",
"Aikaterini Maria",
""
],
[
"Balaskas",
"Konstantinos",
""
],
[
"Zervakis",
"Georgios",
""
],
[
"Amrouch",
"Hussam",
""
],
[
"Anagnostopoulos",
"Iraklis",
""
]
] |
2504.10974 | Zhisheng Zhang | Zhisheng Zhang, Peng Zhang, Fengxiang Wang, Liangli Ma, Fuchun Sun | Self-Supervised Enhancement of Forward-Looking Sonar Images: Bridging Cross-Modal Degradation Gaps through Feature Space Transformation and Multi-Frame Fusion | null | null | null | null | cs.CV eess.IV | http://creativecommons.org/licenses/by/4.0/ | Enhancing forward-looking sonar images is critical for accurate underwater target detection. Current deep learning methods mainly rely on supervised training with simulated data, but the difficulty in obtaining high-quality real-world paired data limits their practical use and generalization. Although self-supervised approaches from remote sensing partially alleviate data shortages, they neglect the cross-modal degradation gap between sonar and remote sensing images. Directly transferring pretrained weights often leads to overly smooth sonar images, detail loss, and insufficient brightness. To address this, we propose a feature-space transformation that maps sonar images from the pixel domain to a robust feature domain, effectively bridging the degradation gap. Additionally, our self-supervised multi-frame fusion strategy leverages complementary inter-frame information to naturally remove speckle noise and enhance target-region brightness. Experiments on three self-collected real-world forward-looking sonar datasets show that our method significantly outperforms existing approaches, effectively suppressing noise, preserving detailed edges, and substantially improving brightness, demonstrating strong potential for underwater target detection applications. | [
{
"version": "v1",
"created": "Tue, 15 Apr 2025 08:34:56 GMT"
},
{
"version": "v2",
"created": "Wed, 16 Apr 2025 15:58:55 GMT"
},
{
"version": "v3",
"created": "Thu, 29 May 2025 07:29:05 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Zhang",
"Zhisheng",
""
],
[
"Zhang",
"Peng",
""
],
[
"Wang",
"Fengxiang",
""
],
[
"Ma",
"Liangli",
""
],
[
"Sun",
"Fuchun",
""
]
] |
2504.11021 | Uddalok Sen | Christoph Haessig, Mehdi Habibi, and Uddalok Sen | Tunable self-emulsification via viscoelastic control of Marangoni-driven interfacial instabilities | null | null | null | null | cond-mat.soft physics.flu-dyn | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Interfacial instabilities in multicomponent fluidic systems are widespread in nature and in industrial processes, yet controlling their dynamics remains a challenge. Here, we present a strategy to actively tune Marangoni-driven self-emulsification at liquid-liquid interfaces by harnessing fluid viscoelasticity. When a water-alcohol droplet spreads on an oil bath, a radial surface tension gradient induced by selective alcohol evaporation drives an interfacial instability, leading to the spontaneous formation of a dense two-dimensional array of "daughter" droplets. We demonstrate that introducing trace amounts of high-molecular-weight polymers, which introduces viscoelasticity, provides a robust means of controlling this process. Increasing viscoelasticity systematically suppresses the instability, resulting in a delayed onset of fragmentation and longer spreading fingers. By combining high-resolution experimental visualization and theoretical analysis, we uncover a quantitative relationship between the polymer concentration and the finger length prior to breakup. These findings establish a predictive framework for designing viscoelastic interfacial materials with programmable dynamic and offer new opportunities for surface-tension-mediated patterning, emulsification, and fluidic control in soft material systems. | [
{
"version": "v1",
"created": "Tue, 15 Apr 2025 09:45:51 GMT"
},
{
"version": "v2",
"created": "Thu, 29 May 2025 12:33:01 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Haessig",
"Christoph",
""
],
[
"Habibi",
"Mehdi",
""
],
[
"Sen",
"Uddalok",
""
]
] |
2504.11471 | Qingyao Zhang | Qingyao Zhang | Vaidya-Reissner-Nordstr\"om Extension On the White-hole Region | 31 pages, 3 figures | null | null | null | gr-qc math-ph math.MP | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | We develop an analytic model that extends classical white hole geometry by incorporating both radiative dynamics and electric charge. Starting from a maximal analytic extension of the Schwarzschild white hole via Kruskal Szekeres coordinates, we introduce a time dependent mass function, representative of outgoing null dust to model evaporation. Building on this foundation, the study then integrates the Reissner-Nordstr\"om framework to obtain a dynamic, charged white hole solution in double null coordinates. In the resulting Vaidya Reissner Nordstr\"om metric, both the Bondi mass and the associated charge decrease monotonically with retarded time, capturing the interplay of radiation and electromagnetic effects. Detailed analysis of horizon behavior reveals how mass loss and charge shedding modify the causal structure, ensuring that energy conditions are preserved and cosmic censorship is maintained. | [
{
"version": "v1",
"created": "Fri, 11 Apr 2025 18:10:05 GMT"
},
{
"version": "v2",
"created": "Thu, 29 May 2025 12:48:02 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Zhang",
"Qingyao",
""
]
] |
2504.11558 | Mete Erdogan | Mete Erdogan, Cengiz Pehlevan, Alper T. Erdogan | Error Broadcast and Decorrelation as a Potential Artificial and Natural Learning Mechanism | null | null | null | null | cs.LG cs.AI | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | We introduce Error Broadcast and Decorrelation (EBD), a novel learning framework for neural networks that addresses credit assignment by directly broadcasting output errors to individual layers, circumventing weight transport of backpropagation. EBD is rigorously grounded in the stochastic orthogonality property of Minimum Mean Square Error estimators. This fundamental principle states that the error of an optimal estimator is orthogonal to functions of the input. Guided by this insight, EBD defines layerwise loss functions that directly penalize correlations between layer activations and output errors, thereby establishing a principled foundation for error broadcasting. This theoretically sound mechanism naturally leads to the experimentally observed three-factor learning rule and integrates with biologically plausible frameworks to enhance performance and plausibility. Numerical experiments demonstrate EBD's competitive or better performance against other error-broadcast methods on benchmark datasets. Our findings establish EBD as an efficient, biologically plausible, and principled alternative for neural network training. | [
{
"version": "v1",
"created": "Tue, 15 Apr 2025 19:00:53 GMT"
},
{
"version": "v2",
"created": "Thu, 29 May 2025 10:44:04 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Erdogan",
"Mete",
""
],
[
"Pehlevan",
"Cengiz",
""
],
[
"Erdogan",
"Alper T.",
""
]
] |
2504.11695 | Isabel Papadimitriou | Isabel Papadimitriou, Huangyuan Su, Thomas Fel, Sham Kakade, Stephanie Gil | Interpreting the linear structure of vision-language model embedding spaces | null | null | null | null | cs.CV cs.MM | http://creativecommons.org/licenses/by/4.0/ | Vision-language models encode images and text in a joint space, minimizing the distance between corresponding image and text pairs. How are language and images organized in this joint space, and how do the models encode meaning and modality? To investigate this, we train and release sparse autoencoders (SAEs) on the embedding spaces of four vision-language models (CLIP, SigLIP, SigLIP2, and AIMv2). SAEs approximate model embeddings as sparse linear combinations of learned directions, or "concepts". We find that, compared to other methods of linear feature learning, SAEs are better at reconstructing the real embeddings, while also able to retain the most sparsity. Retraining SAEs with different seeds or different data diet leads to two findings: the rare, specific concepts captured by the SAEs are liable to change drastically, but we also show that the key commonly-activating concepts extracted by SAEs are remarkably stable across runs. Interestingly, while most concepts are strongly unimodal in activation, we find they are not merely encoding modality per se. Many lie close to - but not entirely within - the subspace defining modality, suggesting that they encode cross-modal semantics despite their unimodal usage. To quantify this bridging behavior, we introduce the Bridge Score, a metric that identifies concept pairs which are both co-activated across aligned image-text inputs and geometrically aligned in the shared space. This reveals that even unimodal concepts can collaborate to support cross-modal integration. We release interactive demos of the SAEs for all models, allowing researchers to explore the organization of the concept spaces. Overall, our findings uncover a sparse linear structure within VLM embedding spaces that is shaped by modality, yet stitched together through latent bridges-offering new insight into how multimodal meaning is constructed. | [
{
"version": "v1",
"created": "Wed, 16 Apr 2025 01:40:06 GMT"
},
{
"version": "v2",
"created": "Thu, 29 May 2025 00:03:59 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Papadimitriou",
"Isabel",
""
],
[
"Su",
"Huangyuan",
""
],
[
"Fel",
"Thomas",
""
],
[
"Kakade",
"Sham",
""
],
[
"Gil",
"Stephanie",
""
]
] |
2504.12919 | Arsim Kastrati | Arsim Kastrati, Haye Hinrichsen | Retarded Causal Set Propagator in 2D Anti de-Sitter Spacetime | 15 pages, 8 figures | null | null | null | gr-qc hep-th | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | We investigate the viability of Causal Set Theory (CST) as a framework for discretizing $(1+1)$-dimensional anti-de Sitter spacetime (AdS$_{1+1}$). In CST, spacetime is modeled as a locally finite, partially ordered set that reflects the causal structure of events. This fundamentally discrete perspective has shown promise in flat spacetime scenarios, while its application to curved geometries is a current field of study. We address this by studying the retarded scalar field propagator on causal sets generated by Poisson sprinkling in AdS$_{1+1}$. We first show the solution of the continuum propagator using the Klein-Gordon equation, solved in terms of geodesic distance. On the discrete side, we employ Shumans path sum method to derive the corresponding propagator, emphasizing that it is valid even in curved manifolds. By performing numerical simulations, we compare the discrete propagator to its continuum counterpart across various curvature scales. Our results show strong agreement and confirm that the causal set propagator accurately captures the effects of curvature without requiring modifications to the flat-space jump amplitudes. These findings affirm the robustness of CST in approximating field propagation in curved spacetimes and strongly support the ability of causal sets to fully capture the geometry of Lorentzian manifolds. | [
{
"version": "v1",
"created": "Thu, 17 Apr 2025 13:08:20 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Kastrati",
"Arsim",
""
],
[
"Hinrichsen",
"Haye",
""
]
] |
2504.13169 | Tsung-Han Wu | Tsung-Han Wu, Heekyung Lee, Jiaxin Ge, Joseph E. Gonzalez, Trevor Darrell, David M. Chan | Generate, but Verify: Reducing Hallucination in Vision-Language Models with Retrospective Resampling | Preprint. Project Page: https://reverse-vlm.github.io | null | null | null | cs.CV | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Vision-Language Models (VLMs) excel at visual understanding but often suffer from visual hallucinations, where they generate descriptions of nonexistent objects, actions, or concepts, posing significant risks in safety-critical applications. Existing hallucination mitigation methods typically follow one of two paradigms: generation adjustment, which modifies decoding behavior to align text with visual inputs, and post-hoc verification, where external models assess and correct outputs. While effective, generation adjustment methods often rely on heuristics and lack correction mechanisms, while post-hoc verification is complicated, typically requiring multiple models and tending to reject outputs rather than refine them. In this work, we introduce REVERSE, a unified framework that integrates hallucination-aware training with on-the-fly self-verification. By leveraging a new hallucination-verification dataset containing over 1.3M semi-synthetic samples, along with a novel inference-time retrospective resampling technique, our approach enables VLMs to both detect hallucinations during generation and dynamically revise those hallucinations. Our evaluations show that REVERSE achieves state-of-the-art hallucination reduction, outperforming the best existing methods by up to 12% on CHAIR-MSCOCO and 34% on HaloQuest. Our dataset, model, and code are available at: https://reverse-vlm.github.io. | [
{
"version": "v1",
"created": "Thu, 17 Apr 2025 17:59:22 GMT"
},
{
"version": "v2",
"created": "Thu, 29 May 2025 02:55:52 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Wu",
"Tsung-Han",
""
],
[
"Lee",
"Heekyung",
""
],
[
"Ge",
"Jiaxin",
""
],
[
"Gonzalez",
"Joseph E.",
""
],
[
"Darrell",
"Trevor",
""
],
[
"Chan",
"David M.",
""
]
] |
2504.14762 | Sahil Rajesh Dhayalkar | Sahil Rajesh Dhayalkar | A Combinatorial Theory of Dropout: Subnetworks, Graph Geometry, and Generalization | 17 pages (9 pages main content and remaining pages are references, appendix which includes 7 figures, proofs and derivations) | null | null | null | cs.LG cs.AI | http://creativecommons.org/licenses/by/4.0/ | We propose a combinatorial and graph-theoretic theory of dropout by modeling training as a random walk over a high-dimensional graph of binary subnetworks. Each node represents a masked version of the network, and dropout induces stochastic traversal across this space. We define a subnetwork contribution score that quantifies generalization and show that it varies smoothly over the graph. Using tools from spectral graph theory, PAC-Bayes analysis, and combinatorics, we prove that generalizing subnetworks form large, connected, low-resistance clusters, and that their number grows exponentially with network width. This reveals dropout as a mechanism for sampling from a robust, structured ensemble of well-generalizing subnetworks with built-in redundancy. Extensive experiments validate every theoretical claim across diverse architectures. Together, our results offer a unified foundation for understanding dropout and suggest new directions for mask-guided regularization and subnetwork optimization. | [
{
"version": "v1",
"created": "Sun, 20 Apr 2025 23:09:20 GMT"
},
{
"version": "v2",
"created": "Wed, 28 May 2025 21:33:01 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Dhayalkar",
"Sahil Rajesh",
""
]
] |
2504.14922 | Kouichi Hagino | K. Hagino, Zehong Liao, S. Yoshida, M. Kimura, and K. Uzawa | Emulating multi-channel scattering based on the eigenvector continuation in the discrete basis formalism | 8 pages, 5 figures | null | null | KUNS-3046 | nucl-th | http://creativecommons.org/licenses/by/4.0/ | We construct an emulator for a multi-channel scattering problem based on the eigenvector continuation. To this end, we employ the Kohn variational principle formulated in the discrete basis formalism. We apply this to one-dimensional scattering problems with a Gaussian barrier. Both for a single-channel and two-channel problems, we demonstrate that the penetration probability as well as the wave functions are well reproduced by the emulator. In particular, the energy dependence of the penetrability in a wider range of energy from well below the barrier to well above the barrier is successfully reproduced. | [
{
"version": "v1",
"created": "Mon, 21 Apr 2025 07:41:59 GMT"
},
{
"version": "v2",
"created": "Thu, 29 May 2025 00:11:08 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Hagino",
"K.",
""
],
[
"Liao",
"Zehong",
""
],
[
"Yoshida",
"S.",
""
],
[
"Kimura",
"M.",
""
],
[
"Uzawa",
"K.",
""
]
] |
2504.14945 | Jianhao Yan | Jianhao Yan, Yafu Li, Zican Hu, Zhi Wang, Ganqu Cui, Xiaoye Qu, Yu Cheng, Yue Zhang | Learning to Reason under Off-Policy Guidance | Work in progress | null | null | null | cs.LG cs.AI cs.CL | http://creativecommons.org/licenses/by/4.0/ | Recent advances in large reasoning models (LRMs) demonstrate that sophisticated behaviors such as multi-step reasoning and self-reflection can emerge via reinforcement learning with verifiable rewards~(\textit{RLVR}). However, existing \textit{RLVR} approaches are inherently ``on-policy'', limiting learning to a model's own outputs and failing to acquire reasoning abilities beyond its initial capabilities. To address this issue, we introduce \textbf{LUFFY} (\textbf{L}earning to reason \textbf{U}nder o\textbf{FF}-polic\textbf{Y} guidance), a framework that augments \textit{RLVR} with off-policy reasoning traces. LUFFY dynamically balances imitation and exploration by combining off-policy demonstrations with on-policy rollouts during training. Specifically, LUFFY combines the Mixed-Policy GRPO framework, which has a theoretically guaranteed convergence rate, alongside policy shaping via regularized importance sampling to avoid superficial and rigid imitation during mixed-policy training. Compared with previous RLVR methods, LUFFY achieves an over \textbf{+6.4} average gain across six math benchmarks and an advantage of over \textbf{+6.2} points in out-of-distribution tasks. Most significantly, we show that LUFFY successfully trains weak models in scenarios where on-policy RLVR completely fails. These results provide compelling evidence that LUFFY transcends the fundamental limitations of on-policy RLVR and demonstrates the great potential of utilizing off-policy guidance in RLVR. | [
{
"version": "v1",
"created": "Mon, 21 Apr 2025 08:09:13 GMT"
},
{
"version": "v2",
"created": "Tue, 22 Apr 2025 15:37:25 GMT"
},
{
"version": "v3",
"created": "Tue, 20 May 2025 07:29:49 GMT"
},
{
"version": "v4",
"created": "Thu, 29 May 2025 06:14:43 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Yan",
"Jianhao",
""
],
[
"Li",
"Yafu",
""
],
[
"Hu",
"Zican",
""
],
[
"Wang",
"Zhi",
""
],
[
"Cui",
"Ganqu",
""
],
[
"Qu",
"Xiaoye",
""
],
[
"Cheng",
"Yu",
""
],
[
"Zhang",
"Yue",
""
]
] |
2504.15046 | Shilin Zhang | Shilin Zhang, Zican Hu, Wenhao Wu, Xinyi Xie, Jianxiang Tang, Chunlin Chen, Daoyi Dong, Yu Cheng, Zhenhong Sun, Zhi Wang | Text-to-Decision Agent: Offline Meta-Reinforcement Learning from Natural Language Supervision | 18 pages, 8 figures | null | null | null | cs.AI | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Offline meta-RL usually tackles generalization by inferring task beliefs from high-quality samples or warmup explorations. The restricted form limits their generality and usability since these supervision signals are expensive and even infeasible to acquire in advance for unseen tasks. Learning directly from the raw text about decision tasks is a promising alternative to leverage a much broader source of supervision. In the paper, we propose \textbf{T}ext-to-\textbf{D}ecision \textbf{A}gent (\textbf{T2DA}), a simple and scalable framework that supervises offline meta-RL with natural language. We first introduce a generalized world model to encode multi-task decision data into a dynamics-aware embedding space. Then, inspired by CLIP, we predict which textual description goes with which decision embedding, effectively bridging their semantic gap via contrastive language-decision pre-training and aligning the text embeddings to comprehend the environment dynamics. After training the text-conditioned generalist policy, the agent can directly realize zero-shot text-to-decision generation in response to language instructions. Comprehensive experiments on MuJoCo and Meta-World benchmarks show that T2DA facilitates high-capacity zero-shot generalization and outperforms various types of baselines. | [
{
"version": "v1",
"created": "Mon, 21 Apr 2025 12:00:20 GMT"
},
{
"version": "v2",
"created": "Tue, 22 Apr 2025 05:56:57 GMT"
},
{
"version": "v3",
"created": "Thu, 29 May 2025 06:17:36 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Zhang",
"Shilin",
""
],
[
"Hu",
"Zican",
""
],
[
"Wu",
"Wenhao",
""
],
[
"Xie",
"Xinyi",
""
],
[
"Tang",
"Jianxiang",
""
],
[
"Chen",
"Chunlin",
""
],
[
"Dong",
"Daoyi",
""
],
[
"Cheng",
"Yu",
""
],
[
"Sun",
"Zhenhong",
""
],
[
"Wang",
"Zhi",
""
]
] |
2504.15319 | Hristu Culetu | Hristu Culetu | On a star with expanding anisotropic fluid | 6 pages, no figures, title and sec.2 modified | null | null | null | gr-qc | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | The generalized Gullstrand-Painleve geometry is investigated for expanding matter. Compared to other studies, we take into account an anisotropic stress tensor as the source of curvature with an equation of state resembling the MIT bag model form. The spacetime is of de Sitter type for $t>>1/\sqrt{\Lambda}, \Lambda$ being the equivalent cosmological constant. The energy density and pressures of the fluid are only time dependent. The radial geodesics are computed. | [
{
"version": "v1",
"created": "Sun, 20 Apr 2025 20:32:18 GMT"
},
{
"version": "v2",
"created": "Wed, 28 May 2025 20:07:54 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Culetu",
"Hristu",
""
]
] |
2504.15966 | Guga Khundzakishvili | Guga Khundzakishvili, Bishnu P. Belbase, Pravin Mahendran, Kevin Zhang, Hanjing Xu, Eliana Stoyanoff, Joseph G. Checkelsky, Yaohua Liu, Linda Ye, Arnab Banerjee | Criticality and magnetic phases of Ising Shastry-Sutherland candidate holmium tetraboride | null | null | 10.3390/ma18112504 | null | cond-mat.str-el cond-mat.mtrl-sci | http://creativecommons.org/licenses/by/4.0/ | Frustrated magnetic systems arising in geometrically constrained lattices represent rich platforms for exploring unconventional phases of matter, including fractional magnetization plateaus, incommensurate orders, and complex domain dynamics. However, determining the microscopic spin configurations that stabilize such phases is a key challenge, especially when in-plane and out-of-plane spin components coexist and compete. Here, we combine neutron scattering and magnetic susceptibility experiments with simulations to investigate the emergence of field-induced fractional plateaus and the related criticality in a frustrated magnet holmium tetraboride (HoB4) that represents the family of rare earth tetraborides that crystalize in a Shastry-Sutherland lattice in the ab plane. We focus on the interplay between classical and quantum criticality near phase boundaries as well as the role of material defects in the stabilization of the ordered phases. We find that simulations using classical annealing can explain certain observed features in the experimental Laue diffraction and the origin of multiple magnetization plateaus. Our results show that defects and out of plane interactions play an important role and can guide the route towards resolving microscopic spin textures in highly frustrated magnets. | [
{
"version": "v1",
"created": "Tue, 22 Apr 2025 15:05:09 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Khundzakishvili",
"Guga",
""
],
[
"Belbase",
"Bishnu P.",
""
],
[
"Mahendran",
"Pravin",
""
],
[
"Zhang",
"Kevin",
""
],
[
"Xu",
"Hanjing",
""
],
[
"Stoyanoff",
"Eliana",
""
],
[
"Checkelsky",
"Joseph G.",
""
],
[
"Liu",
"Yaohua",
""
],
[
"Ye",
"Linda",
""
],
[
"Banerjee",
"Arnab",
""
]
] |
2504.16076 | Claire Lamman | Claire Lamman, Jonathan Blazek, Daniel J. Eisenstein | Optimal intrinsic alignment estimators in the presence of redshift-space distortions | For an accessible summary of this paper, see https://cmlamman.github.io/doc/optimalRSDstats-annotation.pdf | null | null | null | astro-ph.CO | http://creativecommons.org/licenses/by/4.0/ | We present estimators for quantifying intrinsic alignments in large spectroscopic surveys that efficiently capture line-of-sight (LOS) information while being relatively insensitive to redshift-space distortions (RSD). We demonstrate that changing the LOS integration range, {\Pi}max, as a function of transverse separation outperforms the conventional choice of a single {\Pi}max value. This is further improved by replacing the flat {\Pi}max cut with a LOS weighting based on shape projection and RSD. Although these estimators incorporate additional LOS information, they are projected correlations that exhibit signal-to-noise ratios comparable to 3D correlation functions, such as the IA quadrupole. Using simulations from Abacus Summit, we evaluate these estimators and provide recommended {\Pi}max values and weights for projected separations of 1 - 100 Mpc/h. These will improve measurements of intrinsic alignments in large cosmological surveys and the constraints they provide for both weak lensing and direct cosmological applications. | [
{
"version": "v1",
"created": "Tue, 22 Apr 2025 17:55:52 GMT"
},
{
"version": "v2",
"created": "Wed, 28 May 2025 18:32:19 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Lamman",
"Claire",
""
],
[
"Blazek",
"Jonathan",
""
],
[
"Eisenstein",
"Daniel J.",
""
]
] |
2504.16850 | Salar Zamani Salimi | Salar Zamani Salimi, Andrea Gruber, Nicol\`o Scapin, Luca Brandt | Evaporation of Finite-Size Ammonia and n-Heptane Droplets in Weakly Compressible Turbulence: An Interface-Resolved DNS Study | null | null | null | null | physics.flu-dyn | http://creativecommons.org/licenses/by-sa/4.0/ | This study presents direct numerical simulation (DNS) of finite-size, interface-resolved ammonia and n-heptane droplets evaporating in decaying homogeneous isotropic turbulence. Simulations are conducted for each fuel to model the dynamics in a dense spray region, where the liquid volume fraction exceeds $\mathcal{O}(10^{-2})$. The focus is on investigating the complex interactions between droplets, turbulence, and phase change, with emphasis on droplet-droplet interactions and their influence on the evaporation process. The present study also explores how varying turbulence intensities affect the evaporation rates of each fuel, unveiling the differences in the coalescence and energy transfer from the liquid to the gaseous phase. The results reveal that, when comparing ammonia with n-heptane with equal liquid volume fractions, ammonia exhibits faster initial evaporation due to its higher volatility. However, this rate declines over time as frequent droplet coalescence reduces the total surface area available for evaporation. When numerical experiments are initialized with equal energy content, increasing turbulence intensity enhances the evaporation of n-heptane throughout the simulation, while ammonia evaporation soon becomes less sensitive to turbulence due to rapid vapor saturation. These findings are relevant to improving predictive CFD models and optimizing fuel injection in spray-combustion applications, especially under high-pressure conditions. | [
{
"version": "v1",
"created": "Wed, 23 Apr 2025 16:11:29 GMT"
},
{
"version": "v2",
"created": "Thu, 29 May 2025 17:37:45 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Salimi",
"Salar Zamani",
""
],
[
"Gruber",
"Andrea",
""
],
[
"Scapin",
"Nicolò",
""
],
[
"Brandt",
"Luca",
""
]
] |
2504.16857 | Malte Henkel | Malte Henkel | Physical ageing from generalised time-translation-invariance | Latex 2e, 1+56 pages, 6 figures, 4 tables | Nucl. Phys. B1017, 116968 (2025) | 10.1016/j.nuclphysb.2025.116968 | null | cond-mat.stat-mech hep-th math-ph math.MP | http://creativecommons.org/licenses/by/4.0/ | A generalised form of time-translation-invariance permits to re-derive the known generic phenomenology of ageing, which arises in classical many-body systems after a quench from an initially disordered system to a temperature $T\leq T_c$, at or below the critical temperature $T_c$. Generalised time-translation-invariance is obtained, out of equilibrium, from a change of representation of the Lie algebra generators of the dynamical symmetries of scale-invariance and time-translation-invariance. Observable consequences include the algebraic form of the scaling functions for large arguments of the two-time auto-correlators and auto-responses, the equality of the auto-correlation and the auto-response exponents $\lambda_C=\lambda_R$, the cross-over scaling form for an initially magnetised critical system and the explanation of a novel finite-size scaling if the auto-correlator or auto-response converge for large arguments $y=t/s\gg 1$ to a plateau. For global two-time correlators, the time-dependence involving the initial critical slip exponent $\Theta$ is confirmed and is generalised to all temperatures below criticality and to the global two-time response function, and their finite-size scaling is derived as well. This also includes the time-dependence of the squared global order-parameter. The celebrate Janssen-Schaub-Schmittmann scaling relation with the auto-correlation exponent is thereby extended to all temperatures below the critical temperature. A simple criterion on the relevance of non-linear terms in the stochastic equation of motion is derived, taking the dimensionality of couplings into account. Its applicability in a wide class of models is confirmed, for temperatures $T\leq T_c$. Relevance to experiments is also discussed. | [
{
"version": "v1",
"created": "Wed, 23 Apr 2025 16:26:47 GMT"
},
{
"version": "v2",
"created": "Wed, 21 May 2025 09:01:11 GMT"
},
{
"version": "v3",
"created": "Wed, 28 May 2025 09:17:42 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Henkel",
"Malte",
""
]
] |
2504.17502 | Aviv Slobodkin | Aviv Slobodkin, Hagai Taitelbaum, Yonatan Bitton, Brian Gordon, Michal Sokolik, Nitzan Bitton Guetta, Almog Gueta, Royi Rassin, Dani Lischinski, Idan Szpektor | RefVNLI: Towards Scalable Evaluation of Subject-driven Text-to-image Generation | null | null | null | null | cs.CV | http://creativecommons.org/licenses/by/4.0/ | Subject-driven text-to-image (T2I) generation aims to produce images that align with a given textual description, while preserving the visual identity from a referenced subject image. Despite its broad downstream applicability - ranging from enhanced personalization in image generation to consistent character representation in video rendering - progress in this field is limited by the lack of reliable automatic evaluation. Existing methods either assess only one aspect of the task (i.e., textual alignment or subject preservation), misalign with human judgments, or rely on costly API-based evaluation. To address this gap, we introduce RefVNLI, a cost-effective metric that evaluates both textual alignment and subject preservation in a single run. Trained on a large-scale dataset derived from video-reasoning benchmarks and image perturbations, RefVNLI outperforms or statistically matches existing baselines across multiple benchmarks and subject categories (e.g., \emph{Animal}, \emph{Object}), achieving up to 6.4-point gains in textual alignment and 5.9-point gains in subject preservation. | [
{
"version": "v1",
"created": "Thu, 24 Apr 2025 12:44:51 GMT"
},
{
"version": "v2",
"created": "Thu, 29 May 2025 09:50:41 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Slobodkin",
"Aviv",
""
],
[
"Taitelbaum",
"Hagai",
""
],
[
"Bitton",
"Yonatan",
""
],
[
"Gordon",
"Brian",
""
],
[
"Sokolik",
"Michal",
""
],
[
"Guetta",
"Nitzan Bitton",
""
],
[
"Gueta",
"Almog",
""
],
[
"Rassin",
"Royi",
""
],
[
"Lischinski",
"Dani",
""
],
[
"Szpektor",
"Idan",
""
]
] |
2504.18428 | Yiming Wang | Yiming Wang, Pei Zhang, Jialong Tang, Haoran Wei, Baosong Yang, Rui Wang, Chenshu Sun, Feitong Sun, Jiran Zhang, Junxuan Wu, Qiqian Cang, Yichang Zhang, Fei Huang, Junyang Lin, Fei Huang, Jingren Zhou | PolyMath: Evaluating Mathematical Reasoning in Multilingual Contexts | 50 pages, 19 tables, 9 figures | null | null | null | cs.CL | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | In this paper, we introduce PolyMath, a multilingual mathematical reasoning benchmark covering 18 languages and 4 easy-to-hard difficulty levels. Our benchmark ensures difficulty comprehensiveness, language diversity, and high-quality translation, making it a highly discriminative multilingual mathematical benchmark in the era of reasoning LLMs. We conduct a comprehensive evaluation for advanced LLMs and find that even Qwen-3-235B-A22B-Thinking and Gemini-2.5-pro, achieve only 54.6 and 52.2 benchmark scores, with about 40% accuracy under the highest level From a language perspective, our benchmark reveals several key challenges of LLMs in multilingual reasoning: (1) Reasoning performance varies widely across languages for current LLMs; (2) Input-output language consistency is low in reasoning LLMs and may be correlated with performance; (3) The thinking length differs significantly by language for current LLMs. Additionally, we demonstrate that controlling the output language in the instructions has the potential to affect reasoning performance, especially for some low-resource languages, suggesting a promising direction for improving multilingual capabilities in LLMs. | [
{
"version": "v1",
"created": "Fri, 25 Apr 2025 15:39:04 GMT"
},
{
"version": "v2",
"created": "Wed, 30 Apr 2025 13:10:37 GMT"
},
{
"version": "v3",
"created": "Thu, 29 May 2025 08:42:37 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Wang",
"Yiming",
""
],
[
"Zhang",
"Pei",
""
],
[
"Tang",
"Jialong",
""
],
[
"Wei",
"Haoran",
""
],
[
"Yang",
"Baosong",
""
],
[
"Wang",
"Rui",
""
],
[
"Sun",
"Chenshu",
""
],
[
"Sun",
"Feitong",
""
],
[
"Zhang",
"Jiran",
""
],
[
"Wu",
"Junxuan",
""
],
[
"Cang",
"Qiqian",
""
],
[
"Zhang",
"Yichang",
""
],
[
"Huang",
"Fei",
""
],
[
"Lin",
"Junyang",
""
],
[
"Huang",
"Fei",
""
],
[
"Zhou",
"Jingren",
""
]
] |
2504.18630 | Viktor T. Toth | Viktor T. Toth | Peeking under the clouds: Is exoplanet imaging with the solar gravitational lens feasible? | 13 pages, 3 figures. Accepted for publication in RAA | Res. Astron. Astrophys. 25 065007 (2025) | 10.1088/1674-4527/add3a3 | null | astro-ph.IM gr-qc | http://creativecommons.org/licenses/by/4.0/ | Exoplanet imaging using the solar gravitational lens is an enticing prospect. The fundamental physical properties of the lens, including its angular resolution and light amplification, promise exceptional capabilities. These expectations, however, are tempered by the realization of numerous challenges, including imperfections of the lens itself, noise sources, the properties of the imaging target and difficult technical issues. We discuss, in particular, a subject not previously addressed, the impact of temporally varying surface features, notably a variable cloud cover, obscuring the target exoplanet. This has a substantial detrimental effect on image recovery, leading to our cautious assessment of the practical feasibility of using the Sun's gravitational field as an effective telescope. | [
{
"version": "v1",
"created": "Fri, 25 Apr 2025 18:15:03 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Toth",
"Viktor T.",
""
]
] |
2504.19597 | Koji Nishida | Koji Nishida | Depth Sensitivity of Hilbert Coefficients | 17 pages | null | null | null | math.AC math.AG | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | The purpose of this paper is to explain about the depth sensitivity of the Hilbert coefficients defined for finitely generated graded modules over graded rings. The main result generalize the well known fact that the Cohen-Macaulayness of graded modules can be characterized using their multiplicities. | [
{
"version": "v1",
"created": "Mon, 28 Apr 2025 09:00:27 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Nishida",
"Koji",
""
]
] |
2504.21336 | Linshan Wu | Linshan Wu, Yuxiang Nie, Sunan He, Jiaxin Zhuang, Luyang Luo, Neeraj Mahboobani, Varut Vardhanabhuti, Ronald Cheong Kin Chan, Yifan Peng, Pranav Rajpurkar, Hao Chen | UniBiomed: A Universal Foundation Model for Grounded Biomedical Image Interpretation | The first universal foundation model for grounded biomedical image interpretation | null | null | null | cs.CV | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | The integration of AI-assisted biomedical image analysis into clinical practice demands AI-generated findings that are not only accurate but also interpretable to clinicians. However, existing biomedical AI models generally lack the ability to simultaneously generate diagnostic findings and localize corresponding biomedical objects. This limitation makes it challenging for clinicians to correlate AI-generated findings with visual evidence (e.g., tiny lesions) in images and interpret the results of AI models. To address this challenge, we introduce UniBiomed, the first universal foundation model for grounded biomedical image interpretation, which is capable of generating accurate diagnostic findings and simultaneously segmenting the corresponding biomedical targets. UniBiomed is based on a novel integration of Multi-modal Large Language Model and Segment Anything Model, which can effectively unify diverse biomedical tasks in universal training for advancing grounded interpretation. To develop UniBiomed, we curate a large-scale dataset comprising over 27 million triplets of images, region annotations, and text descriptions across ten biomedical imaging modalities. Extensive validation on 70 internal and 14 external datasets demonstrated the state-of-the-art performance of UniBiomed in diverse biomedical tasks, including image segmentation, disease recognition, region-aware diagnosis, vision question answering, and report generation. In summary, UniBiomed is a powerful and versatile biomedical foundation model, unlocking the untapped grounded interpretation capability for optimizing AI-assisted biomedical image analysis. | [
{
"version": "v1",
"created": "Wed, 30 Apr 2025 05:51:48 GMT"
},
{
"version": "v2",
"created": "Thu, 29 May 2025 05:14:48 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Wu",
"Linshan",
""
],
[
"Nie",
"Yuxiang",
""
],
[
"He",
"Sunan",
""
],
[
"Zhuang",
"Jiaxin",
""
],
[
"Luo",
"Luyang",
""
],
[
"Mahboobani",
"Neeraj",
""
],
[
"Vardhanabhuti",
"Varut",
""
],
[
"Chan",
"Ronald Cheong Kin",
""
],
[
"Peng",
"Yifan",
""
],
[
"Rajpurkar",
"Pranav",
""
],
[
"Chen",
"Hao",
""
]
] |
2504.21845 | Abhinav Vaishya | Jefrin Sharmitha Prabhu, Abhinav Vaishya, Shobhit Bhatnagar, Aryaman Manish Kolhe, V. Lalitha and P. Vijay Kumar | On the Efficacy of the Peeling Decoder for the Quantum Expander Code | null | null | null | null | quant-ph cs.IT math.IT | http://creativecommons.org/licenses/by/4.0/ | The problem of recovering from qubit erasures has recently gained attention as erasures occur in many physical systems such as photonic systems, trapped ions, superconducting qubits and circuit quantum electrodynamics. While several linear-time decoders for error correction are known, their error-correcting capability is limited to half the minimum distance of the code, whereas erasure correction allows one to go beyond this limit. As in the classical case, stopping sets pose a major challenge in designing efficient erasure decoders for quantum LDPC codes. In this paper, we show through simulation, that an attractive alternative here, is the use of quantum expander codes in conjunction with the peeling decoder that has linear complexity. We also discuss additional techniques including small-set-flip decoding, that can be applied following the peeling operation, to improve decoding performance and their associated complexity. | [
{
"version": "v1",
"created": "Wed, 30 Apr 2025 17:54:49 GMT"
},
{
"version": "v2",
"created": "Thu, 29 May 2025 16:21:19 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Prabhu",
"Jefrin Sharmitha",
""
],
[
"Vaishya",
"Abhinav",
""
],
[
"Bhatnagar",
"Shobhit",
""
],
[
"Kolhe",
"Aryaman Manish",
""
],
[
"Lalitha",
"V.",
""
],
[
"Kumar",
"P. Vijay",
""
]
] |
2505.00091 | Tengchao Zhang | Tengchao Zhang, Yonglin Tian, Fei Lin, Jun Huang, Patrik P. S\"uli, Rui Qin and Fei-Yue Wang | CoordField: Coordination Field for Agentic UAV Task Allocation In Low-altitude Urban Scenarios | Submitted ITSC 2025 | null | null | null | cs.RO cs.AI | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | With the increasing demand for heterogeneous Unmanned Aerial Vehicle (UAV) swarms to perform complex tasks in urban environments, system design now faces major challenges, including efficient semantic understanding, flexible task planning, and the ability to dynamically adjust coordination strategies in response to evolving environmental conditions and continuously changing task requirements. To address the limitations of existing approaches, this paper proposes coordination field agentic system for coordinating heterogeneous UAV swarms in complex urban scenarios. In this system, large language models (LLMs) is responsible for interpreting high-level human instructions and converting them into executable commands for the UAV swarms, such as patrol and target tracking. Subsequently, a Coordination field mechanism is proposed to guide UAV motion and task selection, enabling decentralized and adaptive allocation of emergent tasks. A total of 50 rounds of comparative testing were conducted across different models in a 2D simulation space to evaluate their performance. Experimental results demonstrate that the proposed system achieves superior performance in terms of task coverage, response time, and adaptability to dynamic changes. | [
{
"version": "v1",
"created": "Wed, 30 Apr 2025 18:02:45 GMT"
},
{
"version": "v2",
"created": "Sat, 03 May 2025 16:45:20 GMT"
},
{
"version": "v3",
"created": "Wed, 28 May 2025 23:31:13 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Zhang",
"Tengchao",
""
],
[
"Tian",
"Yonglin",
""
],
[
"Lin",
"Fei",
""
],
[
"Huang",
"Jun",
""
],
[
"Süli",
"Patrik P.",
""
],
[
"Qin",
"Rui",
""
],
[
"Wang",
"Fei-Yue",
""
]
] |
2505.00288 | Tri P. Nguyen | Tri P. Nguyen, Ilon Joseph, Mayya Tokman | Nystr\"om Type Exponential Integrators for Strongly Magnetized Charged Particle Dynamics | null | null | null | null | physics.comp-ph cs.NA math.NA physics.plasm-ph | http://creativecommons.org/licenses/by-nc-nd/4.0/ | Calculating the dynamics of charged particles in electromagnetic fields (i.e. the particle pushing problem) is one of the most computationally intensive components of particle-in-cell (PIC) methods for plasma physics simulations. This task is especially challenging when the plasma is strongly magnetized, since in this case the particle motion consists of a wide range of temporal scales from highly oscillatory fast gyromotion to slow macroscopic behavior and the resulting numerical model is very stiff. Current state-of-the-art time integrators used to simulate particle motion have limitations given the severe numerical stiffness of the problem and more efficient methods are of interest. Recently, exponential integrators have been proposed as a promising new approach for these simulations and shown to offer computational advantages over commonly used schemes. Exponential methods can solve linear problems exactly and are $A$-stable. In this paper, the standard exponential algorithms framework is extended to derive Nystr\"om-type exponential methods that integrate the Newtonian equations of motion as a second-order differential equation. Specific Nystr\"om-type schemes of second and third orders are derived and applied to strongly magnetized particle pushing problems. Numerical experiments are presented to demonstrate that the Nystr\"om-type exponential integrators can provide significant improvement in computational efficiency over the standard exponential methods. | [
{
"version": "v1",
"created": "Thu, 01 May 2025 04:22:49 GMT"
},
{
"version": "v2",
"created": "Wed, 14 May 2025 01:56:48 GMT"
},
{
"version": "v3",
"created": "Thu, 29 May 2025 03:38:24 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Nguyen",
"Tri P.",
""
],
[
"Joseph",
"Ilon",
""
],
[
"Tokman",
"Mayya",
""
]
] |
2505.00427 | Rhea Alexander | Rhea Alexander | A new approximate Eastin-Knill theorem | 20 pages, 2 figures. Comments welcome! | null | null | null | quant-ph | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Transversal encoded gatesets are highly desirable for fault tolerant quantum computing. However, a quantum error correcting code which exactly corrects for local erasure noise and supports a universal set of transversal gates is ruled out by the Eastin-Knill theorem. Here we provide a new approximate Eastin-Knill theorem for the single-shot regime when we allow for some probability of error in the decoding. In particular, we show that a quantum error correcting code can support a universal set of transversal gates and approximately correct for local erasure if and only if the conditional min-entropy of the Choi state of the encoding and noise channel is upper bounded by a simple function of the worst-case error probability. Our no-go theorem can be computed by solving a semidefinite program, and, in the spirit of the original Eastin-Knill theorem, is formulated in terms of a condition that is both necessary and sufficient, ensuring achievability whenever it is passed. As an example, we find that with $n=100$ physical qutrits we can encode $k=1$ logical qubit in the $W$-state code, which admits a universal transversal set of gates and corrects for single subsystem erasure with error probability of $\varepsilon = 0.005$. To establish our no-go result, we leverage tools from the resource theory of asymmetry, where, in the single-shot regime, a single (output state-dependent) resource monotone governs all state purifications. | [
{
"version": "v1",
"created": "Thu, 01 May 2025 09:56:26 GMT"
},
{
"version": "v2",
"created": "Thu, 29 May 2025 10:03:49 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Alexander",
"Rhea",
""
]
] |
2505.00612 | Megan Risdal | D. Sculley, Will Cukierski, Phil Culliton, Sohier Dane, Maggie Demkin, Ryan Holbrook, Addison Howard, Paul Mooney, Walter Reade, Megan Risdal and Nate Keating | Position: AI Competitions Provide the Gold Standard for Empirical Rigor in GenAI Evaluation | null | null | null | null | cs.AI | http://creativecommons.org/licenses/by/4.0/ | In this position paper, we observe that empirical evaluation in Generative AI is at a crisis point since traditional ML evaluation and benchmarking strategies are insufficient to meet the needs of evaluating modern GenAI models and systems. There are many reasons for this, including the fact that these models typically have nearly unbounded input and output spaces, typically do not have a well defined ground truth target, and typically exhibit strong feedback loops and prediction dependence based on context of previous model outputs. On top of these critical issues, we argue that the problems of leakage and contamination are in fact the most important and difficult issues to address for GenAI evaluations. Interestingly, the field of AI Competitions has developed effective measures and practices to combat leakage for the purpose of counteracting cheating by bad actors within a competition setting. This makes AI Competitions an especially valuable (but underutilized) resource. Now is time for the field to view AI Competitions as the gold standard for empirical rigor in GenAI evaluation, and to harness and harvest their results with according value. | [
{
"version": "v1",
"created": "Thu, 01 May 2025 15:43:51 GMT"
},
{
"version": "v2",
"created": "Thu, 29 May 2025 01:48:23 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Sculley",
"D.",
""
],
[
"Cukierski",
"Will",
""
],
[
"Culliton",
"Phil",
""
],
[
"Dane",
"Sohier",
""
],
[
"Demkin",
"Maggie",
""
],
[
"Holbrook",
"Ryan",
""
],
[
"Howard",
"Addison",
""
],
[
"Mooney",
"Paul",
""
],
[
"Reade",
"Walter",
""
],
[
"Risdal",
"Megan",
""
],
[
"Keating",
"Nate",
""
]
] |
2505.00926 | Ruiquan Huang | Ruiquan Huang, Yingbin Liang, Jing Yang | How Transformers Learn Regular Language Recognition: A Theoretical Study on Training Dynamics and Implicit Bias | accepted by ICML 2025 | null | null | null | cs.LG cs.CL stat.ML | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Language recognition tasks are fundamental in natural language processing (NLP) and have been widely used to benchmark the performance of large language models (LLMs). These tasks also play a crucial role in explaining the working mechanisms of transformers. In this work, we focus on two representative tasks in the category of regular language recognition, known as `even pairs' and `parity check', the aim of which is to determine whether the occurrences of certain subsequences in a given sequence are even. Our goal is to explore how a one-layer transformer, consisting of an attention layer followed by a linear layer, learns to solve these tasks by theoretically analyzing its training dynamics under gradient descent. While even pairs can be solved directly by a one-layer transformer, parity check need to be solved by integrating Chain-of-Thought (CoT), either into the inference stage of a transformer well-trained for the even pairs task, or into the training of a one-layer transformer. For both problems, our analysis shows that the joint training of attention and linear layers exhibits two distinct phases. In the first phase, the attention layer grows rapidly, mapping data sequences into separable vectors. In the second phase, the attention layer becomes stable, while the linear layer grows logarithmically and approaches in direction to a max-margin hyperplane that correctly separates the attention layer outputs into positive and negative samples, and the loss decreases at a rate of $O(1/t)$. Our experiments validate those theoretical results. | [
{
"version": "v1",
"created": "Fri, 02 May 2025 00:07:35 GMT"
},
{
"version": "v2",
"created": "Mon, 05 May 2025 22:08:55 GMT"
},
{
"version": "v3",
"created": "Wed, 28 May 2025 23:17:46 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Huang",
"Ruiquan",
""
],
[
"Liang",
"Yingbin",
""
],
[
"Yang",
"Jing",
""
]
] |
2505.00971 | Kumpei Imamura | Kumpei Imamura, Ryuichi Namba, Riku Ishioka, Kota Ishihara, Yuji Matsuda, Seunghun Lee, Eun-Gook Moon, Kenichiro Hashimoto, Takasada Shibauchi | Bulk excitations in ultraclean $\alpha$-RuCl$_3$: Quantitative evidence for Majorana dispersions in a Kitaev quantum spin liquid | 6 pages, 4 figures | Phys. Rev. B 111, 184440 (2025) | 10.1103/PhysRevB.111.184440 | null | cond-mat.str-el | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | The spin-orbit coupled Mott insulator $\alpha$-RuCl$_3$ has emerged as a prime candidate for realizing the Kitaev quantum spin liquid (KQSL), characterized by Majorana quasiparticles, whose edge states exhibit a distinctive half-integer quantized thermal Hall conductivity. However, its van der Waals nature makes its thermal Hall response highly sensitive to structural disorder, leading to sample-dependent variations. Here, we investigate low-energy bulk excitations in the field-induced quantum disordered (FIQD) state of newly available ultraclean single crystals of $\alpha$-RuCl$_3$. High-resolution specific heat measurements under in-plane magnetic field rotation reveal an anisotropic excitation gap, whose field dependence is consistent with the Majorana gap in the KQSL state. Remarkably, when the field aligns with Ru-Ru bond directions, we observe gapless excitations with Dirac-like dispersions that quantitatively match theoretical predictions of Majorana bands based on the reported Kitaev interactions. Our findings in these ultraclean crystals provide strong evidence that the FIQD state of $\alpha$-RuCl$_3$ is a robust KQSL, resilient against small disorder perturbations. | [
{
"version": "v1",
"created": "Fri, 02 May 2025 03:12:41 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Imamura",
"Kumpei",
""
],
[
"Namba",
"Ryuichi",
""
],
[
"Ishioka",
"Riku",
""
],
[
"Ishihara",
"Kota",
""
],
[
"Matsuda",
"Yuji",
""
],
[
"Lee",
"Seunghun",
""
],
[
"Moon",
"Eun-Gook",
""
],
[
"Hashimoto",
"Kenichiro",
""
],
[
"Shibauchi",
"Takasada",
""
]
] |
2505.01792 | Jun-Jie Zhang | Jun-Jie Zhang, Boris I. Yakobson and Shuai Dong | Landau theory description of autferroicity | null | Phys. Rev. Lett. 134, 216801 (2025) | 10.1103/PhysRevLett.134.216801 | null | cond-mat.mtrl-sci cond-mat.str-el | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Autferroics, recently proposed as a sister branch of multiferroics, exhibit strong intrinsic magnetoelectricity, but ferroelectricity and magnetism are mutually exclusive rather than coexisting. Here, a general model is considered based on the Landau theory, to clarify the distinction between multi and autferroics by qualitative change-rotation in Landau free energy landscape and in particular phase mapping. The TiGeSe$_3$ exemplifies a factual material, whose first-principles computed Landau coefficients predict its autferroicity. Our investigations pave the way for an alternative avenue in the pursuit of intrinsically strong magnetoelectrics. | [
{
"version": "v1",
"created": "Sat, 03 May 2025 11:41:19 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Zhang",
"Jun-Jie",
""
],
[
"Yakobson",
"Boris I.",
""
],
[
"Dong",
"Shuai",
""
]
] |
2505.02093 | Grigoriy Shutov | Grigoriy Shutov, Viktor Duplyakov, Shadfar Davoodi, Anton Morozov, Dmitriy Popkov, Kirill Pavlenko, Albert Vainshtein, Viktor Kotezhekov, Sergey Kaygorodov, Boris Belozerov, Mars M Khasanov, Vladimir Vanovskiy, Andrei Osiptsov, Evgeny Burnaev | A Deep Learning-Aided Approach for Estimating Field Permeability Map by Fusing Well Logs, Well Tests, and Seismic Data | null | null | null | null | cs.CE | http://creativecommons.org/licenses/by-nc-nd/4.0/ | Obtaining reliable permeability maps of oil reservoirs is crucial for building a robust and accurate reservoir simulation model and, therefore, designing effective recovery strategies. This problem, however, remains challenging, as it requires the integration of various data sources by experts from different disciplines. Moreover, there are no sources to provide direct information about the inter-well space. In this work, a new method based on the data-fusion approach is proposed for predicting two-dimensional permeability maps on the whole reservoir area. This method utilizes non-parametric regression with a custom kernel shape accounting for different data sources: well logs, well tests, and seismics. A convolutional neural network is developed to process seismic data and then incorporate it with other sources. A multi-stage data fusion procedure helps to artificially increase the training dataset for the seismic interpretation model and finally to construct the adequate permeability map. The proposed methodology of permeability map construction from different sources was tested on a real oil reservoir located in Western Siberia. The results demonstrate that the developed map perfectly corresponds to the permeability estimations in the wells, and the inter-well space permeability predictions are considerably improved through the incorporation of the seismic data. | [
{
"version": "v1",
"created": "Sun, 04 May 2025 12:59:19 GMT"
},
{
"version": "v2",
"created": "Thu, 29 May 2025 07:30:17 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Shutov",
"Grigoriy",
""
],
[
"Duplyakov",
"Viktor",
""
],
[
"Davoodi",
"Shadfar",
""
],
[
"Morozov",
"Anton",
""
],
[
"Popkov",
"Dmitriy",
""
],
[
"Pavlenko",
"Kirill",
""
],
[
"Vainshtein",
"Albert",
""
],
[
"Kotezhekov",
"Viktor",
""
],
[
"Kaygorodov",
"Sergey",
""
],
[
"Belozerov",
"Boris",
""
],
[
"Khasanov",
"Mars M",
""
],
[
"Vanovskiy",
"Vladimir",
""
],
[
"Osiptsov",
"Andrei",
""
],
[
"Burnaev",
"Evgeny",
""
]
] |
2505.02179 | Tao Zhu | Tao Zhu, Qi Yu, Xinru Dong, Shiyu Li, Yue Liu, Jinlong Jiang, Lei Shu | ProDisc-VAD: An Efficient System for Weakly-Supervised Anomaly Detection in Video Surveillance Applications | A newly identified systematic error in our data processing pipeline has affected the calculation and reporting of AUC metrics (notably in Tables [1, 2]). This significantly impacts our main experimental results and conclusions, compromising their reliability. To ensure academic rigor and prevent misleading information, this manuscript is withdrawn for thorough correction and re-evaluation | null | null | null | cs.CV | http://creativecommons.org/licenses/by/4.0/ | Weakly-supervised video anomaly detection (WS-VAD) using Multiple Instance Learning (MIL) suffers from label ambiguity, hindering discriminative feature learning. We propose ProDisc-VAD, an efficient framework tackling this via two synergistic components. The Prototype Interaction Layer (PIL) provides controlled normality modeling using a small set of learnable prototypes, establishing a robust baseline without being overwhelmed by dominant normal data. The Pseudo-Instance Discriminative Enhancement (PIDE) loss boosts separability by applying targeted contrastive learning exclusively to the most reliable extreme-scoring instances (highest/lowest scores). ProDisc-VAD achieves strong AUCs (97.98% ShanghaiTech, 87.12% UCF-Crime) using only 0.4M parameters, over 800x fewer than recent ViT-based methods like VadCLIP, demonstrating exceptional efficiency alongside state-of-the-art performance. Code is available at https://github.com/modadundun/ProDisc-VAD. | [
{
"version": "v1",
"created": "Sun, 04 May 2025 16:42:15 GMT"
},
{
"version": "v2",
"created": "Thu, 29 May 2025 14:18:43 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Zhu",
"Tao",
""
],
[
"Yu",
"Qi",
""
],
[
"Dong",
"Xinru",
""
],
[
"Li",
"Shiyu",
""
],
[
"Liu",
"Yue",
""
],
[
"Jiang",
"Jinlong",
""
],
[
"Shu",
"Lei",
""
]
] |
2505.02322 | Runquan Gui | Runquan Gui, Zhihai Wang, Jie Wang, Chi Ma, Huiling Zhen, Mingxuan Yuan, Jianye Hao, Defu Lian, Enhong Chen, Feng Wu | HyperTree Planning: Enhancing LLM Reasoning via Hierarchical Thinking | arXiv admin note: text overlap with arXiv:2406.14228 by other authors | null | null | null | cs.AI | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Recent advancements have significantly enhanced the performance of large language models (LLMs) in tackling complex reasoning tasks, achieving notable success in domains like mathematical and logical reasoning. However, these methods encounter challenges with complex planning tasks, primarily due to extended reasoning steps, diverse constraints, and the challenge of handling multiple distinct sub-tasks. To address these challenges, we propose HyperTree Planning (HTP), a novel reasoning paradigm that constructs hypertree-structured planning outlines for effective planning. The hypertree structure enables LLMs to engage in hierarchical thinking by flexibly employing the divide-and-conquer strategy, effectively breaking down intricate reasoning steps, accommodating diverse constraints, and managing multiple distinct sub-tasks in a well-organized manner. We further introduce an autonomous planning framework that completes the planning process by iteratively refining and expanding the hypertree-structured planning outlines. Experiments demonstrate the effectiveness of HTP, achieving state-of-the-art accuracy on the TravelPlanner benchmark with Gemini-1.5-Pro, resulting in a 3.6 times performance improvement over o1-preview. | [
{
"version": "v1",
"created": "Mon, 05 May 2025 02:38:58 GMT"
},
{
"version": "v2",
"created": "Thu, 29 May 2025 03:57:01 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Gui",
"Runquan",
""
],
[
"Wang",
"Zhihai",
""
],
[
"Wang",
"Jie",
""
],
[
"Ma",
"Chi",
""
],
[
"Zhen",
"Huiling",
""
],
[
"Yuan",
"Mingxuan",
""
],
[
"Hao",
"Jianye",
""
],
[
"Lian",
"Defu",
""
],
[
"Chen",
"Enhong",
""
],
[
"Wu",
"Feng",
""
]
] |
2505.02488 | Bob Oliver | Bob Oliver | Limits over orbit categories of locally finite groups | null | null | null | null | math.GR math.CT | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | We correct an error in the paper [BLO3], and take the opportunity to examine in more detail the derived functors of inverse limits over orbit categories of (infinite) locally finite groups. The main results show how to reduce this in many cases to limits over orbit categories of finite groups, but we also look at generalizations of the Lyndon-Hochschild-Serre spectral sequence for higher limits over orbit categories for an extension of locally finite groups. | [
{
"version": "v1",
"created": "Mon, 05 May 2025 09:13:34 GMT"
},
{
"version": "v2",
"created": "Thu, 29 May 2025 07:40:26 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Oliver",
"Bob",
""
]
] |
2505.02977 | Tianyu Liang | Tianyu Liang, Chao Chen, Yotam Yaniv, Hengrui Luo, David Tench, Xiaoye S. Li, Aydin Buluc, James Demmel | Parallel GPU-Accelerated Randomized Construction of Approximate Cholesky Preconditioners | null | null | null | null | cs.DC cs.DS cs.NA math.NA | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | We introduce a parallel algorithm to construct a preconditioner for solving a large, sparse linear system where the coefficient matrix is a Laplacian matrix (a.k.a., graph Laplacian). Such a linear system arises from applications such as discretization of a partial differential equation, spectral graph partitioning, and learning problems on graphs. The preconditioner belongs to the family of incomplete factorizations and is purely algebraic. Unlike traditional incomplete factorizations, the new method employs randomization to determine whether or not to keep fill-ins, i.e., newly generated nonzero elements during Gaussian elimination. Since the sparsity pattern of the randomized factorization is unknown, computing such a factorization in parallel is extremely challenging, especially on many-core architectures such as GPUs. Our parallel algorithm dynamically computes the dependency among row/column indices of the Laplacian matrix to be factorized and processes the independent indices in parallel. Furthermore, unlike previous approaches, our method requires little pre-processing time. We implemented the parallel algorithm for multi-core CPUs and GPUs, and we compare their performance to other state-of-the-art methods. | [
{
"version": "v1",
"created": "Mon, 05 May 2025 19:06:45 GMT"
},
{
"version": "v2",
"created": "Thu, 29 May 2025 17:19:10 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Liang",
"Tianyu",
""
],
[
"Chen",
"Chao",
""
],
[
"Yaniv",
"Yotam",
""
],
[
"Luo",
"Hengrui",
""
],
[
"Tench",
"David",
""
],
[
"Li",
"Xiaoye S.",
""
],
[
"Buluc",
"Aydin",
""
],
[
"Demmel",
"James",
""
]
] |
2505.03475 | Zirui Liu | Zirui Liu, Jiatong Li, Yan Zhuang, Qi Liu, Shuanghong Shen, Jie Ouyang, Mingyue Cheng, Shijin Wang | am-ELO: A Stable Framework for Arena-based LLM Evaluation | ICML2025 Accepted | null | null | null | cs.AI cs.LG | http://creativecommons.org/licenses/by/4.0/ | Arena-based evaluation is a fundamental yet significant evaluation paradigm for modern AI models, especially large language models (LLMs). Existing framework based on ELO rating system suffers from the inevitable instability problem due to ranking inconsistency and the lack of attention to the varying abilities of annotators. In this paper, we introduce a novel stable arena framework to address these issues by enhancing the ELO Rating System. Specifically, we replace the iterative update method with a Maximum Likelihood Estimation (MLE) approach, m-ELO, and provide theoretical proof of the consistency and stability of the MLE approach for model ranking. Additionally, we proposed the am-ELO, which modify the Elo Rating's probability function to incorporate annotator abilities, enabling the simultaneous estimation of model scores and annotator reliability. Experiments demonstrate that this method ensures stability, proving that this framework offers a more robust, accurate, and stable evaluation method for LLMs. | [
{
"version": "v1",
"created": "Tue, 06 May 2025 12:28:50 GMT"
},
{
"version": "v2",
"created": "Thu, 29 May 2025 08:23:44 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Liu",
"Zirui",
""
],
[
"Li",
"Jiatong",
""
],
[
"Zhuang",
"Yan",
""
],
[
"Liu",
"Qi",
""
],
[
"Shen",
"Shuanghong",
""
],
[
"Ouyang",
"Jie",
""
],
[
"Cheng",
"Mingyue",
""
],
[
"Wang",
"Shijin",
""
]
] |
2505.03551 | Mark Everitt | Mark J. Everitt | From Mass-Shell Factorisation to Spin: An Attempt at a Matrix-Valued Liouville Framework for Relativistic Classical and Quantum Phase-Spacetime | Draft - Fixed serious issue with a previous version of this work and leave the discussion of defamation quantisation into the main text | null | null | null | quant-ph physics.class-ph | http://creativecommons.org/licenses/by/4.0/ | While Liouville's theorem is first-order in time for the phase-space distribution itself, the relativistic mass-shell constraint $p^\mu p_\mu = m^2$ is naively second-order in energy. We argue that it is reasonable to unify both energy branches within a single Hamiltonian by factorizing $(p^2 - m^2)$ in analogy with Dirac's approach in relativistic quantum mechanics. We show the resulting matrix-based Liouville equation remains first order and naturally yields a $4\times4$ matrix-valued probability density function in phase space as a classical analogue of a relativistic spin-half Wigner function. We investigate its classical physics and deformation quantisation. | [
{
"version": "v1",
"created": "Tue, 06 May 2025 14:02:52 GMT"
},
{
"version": "v2",
"created": "Thu, 15 May 2025 07:44:36 GMT"
},
{
"version": "v3",
"created": "Thu, 29 May 2025 10:41:12 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Everitt",
"Mark J.",
""
]
] |
2505.03793 | Xinyue Zeng | Xinyue Zeng, Haohui Wang, Junhong Lin, Jun Wu, Tyler Cody, Dawei Zhou | LENSLLM: Unveiling Fine-Tuning Dynamics for LLM Selection | Accepted by ICML'2025 | null | null | null | cs.LG cs.AI | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | The proliferation of open-sourced Large Language Models (LLMs) and diverse downstream tasks necessitates efficient model selection, given the impracticality of fine-tuning all candidates due to computational constraints. Despite the recent advances in LLM selection, a fundamental research question largely remains nascent: how can we model the dynamic behaviors of LLMs during fine-tuning, thereby enhancing our understanding of their generalization performance across diverse downstream tasks? In this work, we propose a novel theoretical framework that provides a proper lens to assess the generalization capabilities of LLMs, thereby enabling accurate and efficient LLM selection for downstream applications. In particular, we first derive a PAC-Bayesian Generalization Bound that unveils fine-tuning dynamics of LLMs and then introduce LENSLLM, a Neural Tangent Kernel (NTK)-based Rectified Scaling Model that enables accurate performance predictions across diverse tasks while maintaining computational efficiency. Extensive empirical results on 3 large-scale benchmarks demonstrate that our model achieves up to 91.1% accuracy and reduces up to 88.5% computational cost in LLM selection, outperforming 5 state-of-the-art methods. We open-source our proposed LENSLLM model and corresponding results at LensLLM.io. | [
{
"version": "v1",
"created": "Thu, 01 May 2025 15:07:32 GMT"
},
{
"version": "v2",
"created": "Thu, 29 May 2025 14:15:49 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Zeng",
"Xinyue",
""
],
[
"Wang",
"Haohui",
""
],
[
"Lin",
"Junhong",
""
],
[
"Wu",
"Jun",
""
],
[
"Cody",
"Tyler",
""
],
[
"Zhou",
"Dawei",
""
]
] |
2505.03834 | Samuel E. Wurzel | Samuel E. Wurzel, Scott C. Hsu | Update: Progress toward fusion energy breakeven and gain as measured against the Lawson criteria | 14 pages, 4 figures, 4 tables. We welcome additional peer-reviewed results and exact dates (day, month, and year) that experiments occurred for which only the year is listed in the tables. Please email the first author at the provided email address. This replacement includes newly reported data from NIF, a new section and data from JET, and other minor corrections and updates | null | null | null | physics.plasm-ph physics.soc-ph | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | This paper is an update to our earlier paper ''Progress toward fusion energy breakeven and gain as measured against the Lawson criterion'' [Phys. Plasmas 29, 062103 (2022)]. Plots of Lawson parameter and triple product vs. ion temperature and triple product vs. date achieved are updated with recently published experimental results. A new plot of scientific energy gain vs. date achieved is included. Additionally, notes on new experimental results, clarifications, and a correction are included. | [
{
"version": "v1",
"created": "Sun, 04 May 2025 17:35:27 GMT"
},
{
"version": "v2",
"created": "Thu, 29 May 2025 15:07:41 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Wurzel",
"Samuel E.",
""
],
[
"Hsu",
"Scott C.",
""
]
] |
2505.04053 | Masahiro Fukuda | Masahiro Fukuda, Masato Senami, Yoshiaki Sugimoto, and Taisuke Ozaki | Regional chemical potential analysis for material surfaces | 18 pages, 11 figures | null | null | null | cond-mat.mtrl-sci physics.chem-ph | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | We propose a local regional chemical potential (RCP) analysis method based on an energy window scheme to quantitatively estimate the selectivity of atomic and molecular adsorption on surfaces, as well as the strength of chemical bonding forces between a probe tip and a surface in atomic force microscopy (AFM) measurements. In particular, focusing on the local picture of covalent bonding, we use a simple H$_2$ molecular model to demonstrate a clear relationship between chemical bonding forces and the local RCP. Moreover, density functional theory calculations on molecular systems and diamond C(001) surfaces reveal that the local RCP at the surfaces successfully visualizes electron-donating regions such as dangling bonds and double bonds. These results suggest that the local RCP can serve as an effective measure to analyze high-resolution non-contact or near-contact AFM images enhanced by chemical bonding forces. | [
{
"version": "v1",
"created": "Wed, 07 May 2025 01:48:03 GMT"
},
{
"version": "v2",
"created": "Thu, 29 May 2025 06:40:11 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Fukuda",
"Masahiro",
""
],
[
"Senami",
"Masato",
""
],
[
"Sugimoto",
"Yoshiaki",
""
],
[
"Ozaki",
"Taisuke",
""
]
] |
2505.04209 | Soumik Dey | Soumik Dey, Hansi Wu, Binbin Li | To Judge or not to Judge: Using LLM Judgements for Advertiser Keyphrase Relevance at eBay | null | null | null | null | cs.IR cs.AI cs.LG | http://creativecommons.org/licenses/by-nc-nd/4.0/ | E-commerce sellers are recommended keyphrases based on their inventory on which they advertise to increase buyer engagement (clicks/sales). The relevance of advertiser keyphrases plays an important role in preventing the inundation of search systems with numerous irrelevant items that compete for attention in auctions, in addition to maintaining a healthy seller perception. In this work, we describe the shortcomings of training Advertiser keyphrase relevance filter models on click/sales/search relevance signals and the importance of aligning with human judgment, as sellers have the power to adopt or reject said keyphrase recommendations. In this study, we frame Advertiser keyphrase relevance as a complex interaction between 3 dynamical systems -- seller judgment, which influences seller adoption of our product, Advertising, which provides the keyphrases to bid on, and Search, who holds the auctions for the same keyphrases. This study discusses the practicalities of using human judgment via a case study at eBay Advertising and demonstrate that using LLM-as-a-judge en-masse as a scalable proxy for seller judgment to train our relevance models achieves a better harmony across the three systems -- provided that they are bound by a meticulous evaluation framework grounded in business metrics. | [
{
"version": "v1",
"created": "Wed, 07 May 2025 08:03:25 GMT"
},
{
"version": "v2",
"created": "Thu, 29 May 2025 05:39:34 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Dey",
"Soumik",
""
],
[
"Wu",
"Hansi",
""
],
[
"Li",
"Binbin",
""
]
] |
2505.04388 | Dario Garcia-Gasulla PhD | Dario Garcia-Gasulla, Jordi Bayarri-Planas, Ashwin Kumar Gururajan, Enrique Lopez-Cuena, Adrian Tormos, Daniel Hinjos, Pablo Bernabeu-Perez, Anna Arias-Duart, Pablo Agustin Martin-Torres, Marta Gonzalez-Mallo, Sergio Alvarez-Napagao, Eduard Ayguad\'e-Parra, Ulises Cort\'es | The Aloe Family Recipe for Open and Specialized Healthcare LLMs | Follow-up work from arXiv:2405.01886 | null | null | null | cs.CL cs.AI | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Purpose: With advancements in Large Language Models (LLMs) for healthcare, the need arises for competitive open-source models to protect the public interest. This work contributes to the field of open medical LLMs by optimizing key stages of data preprocessing and training, while showing how to improve model safety (through DPO) and efficacy (through RAG). The evaluation methodology used, which includes four different types of tests, defines a new standard for the field. The resultant models, shown to be competitive with the best private alternatives, are released with a permisive license.
Methods: Building on top of strong base models like Llama 3.1 and Qwen 2.5, Aloe Beta uses a custom dataset to enhance public data with synthetic Chain of Thought examples. The models undergo alignment with Direct Preference Optimization, emphasizing ethical and policy-aligned performance in the presence of jailbreaking attacks. Evaluation includes close-ended, open-ended, safety and human assessments, to maximize the reliability of results.
Results: Recommendations are made across the entire pipeline, backed by the solid performance of the Aloe Family. These models deliver competitive performance across healthcare benchmarks and medical fields, and are often preferred by healthcare professionals. On bias and toxicity, the Aloe Beta models significantly improve safety, showing resilience to unseen jailbreaking attacks. For a responsible release, a detailed risk assessment specific to healthcare is attached to the Aloe Family models.
Conclusion: The Aloe Beta models, and the recipe that leads to them, are a significant contribution to the open-source medical LLM field, offering top-of-the-line performance while maintaining high ethical requirements. This work sets a new standard for developing and reporting aligned LLMs in healthcare. | [
{
"version": "v1",
"created": "Wed, 07 May 2025 13:13:14 GMT"
},
{
"version": "v2",
"created": "Wed, 28 May 2025 20:14:44 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Garcia-Gasulla",
"Dario",
""
],
[
"Bayarri-Planas",
"Jordi",
""
],
[
"Gururajan",
"Ashwin Kumar",
""
],
[
"Lopez-Cuena",
"Enrique",
""
],
[
"Tormos",
"Adrian",
""
],
[
"Hinjos",
"Daniel",
""
],
[
"Bernabeu-Perez",
"Pablo",
""
],
[
"Arias-Duart",
"Anna",
""
],
[
"Martin-Torres",
"Pablo Agustin",
""
],
[
"Gonzalez-Mallo",
"Marta",
""
],
[
"Alvarez-Napagao",
"Sergio",
""
],
[
"Ayguadé-Parra",
"Eduard",
""
],
[
"Cortés",
"Ulises",
""
]
] |
2505.05034 | Wei Chen | Wei Chen and Shigui Li and Jiacheng Li and Junmei Yang and John Paisley and Delu Zeng | Dequantified Diffusion-Schr{\"o}dinger Bridge for Density Ratio Estimation | null | ICML 2025: Proceedings of the 42nd International Conference on Machine Learning, 2025 | null | null | cs.LG stat.ML | http://creativecommons.org/licenses/by/4.0/ | Density ratio estimation is fundamental to tasks involving f-divergences, yet existing methods often fail under significantly different distributions or inadequately overlapping supports -- the density-chasm and the support-chasm problems. Additionally, prior approaches yield divergent time scores near boundaries, leading to instability. We design $\textbf{D}^3\textbf{RE}$, a unified framework for robust, stable and efficient density ratio estimation. We propose the dequantified diffusion bridge interpolant (DDBI), which expands support coverage and stabilizes time scores via diffusion bridges and Gaussian dequantization. Building on DDBI, the proposed dequantified Schr{\"o}dinger bridge interpolant (DSBI) incorporates optimal transport to solve the Schr{\"o}dinger bridge problem, enhancing accuracy and efficiency. Our method offers uniform approximation and bounded time scores in theory, and outperforms baselines empirically in mutual information and density estimation tasks. | [
{
"version": "v1",
"created": "Thu, 08 May 2025 08:12:16 GMT"
},
{
"version": "v2",
"created": "Mon, 19 May 2025 10:46:33 GMT"
},
{
"version": "v3",
"created": "Thu, 29 May 2025 07:27:43 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Chen",
"Wei",
""
],
[
"Li",
"Shigui",
""
],
[
"Li",
"Jiacheng",
""
],
[
"Yang",
"Junmei",
""
],
[
"Paisley",
"John",
""
],
[
"Zeng",
"Delu",
""
]
] |
2505.05128 | Sujoy Basak | Sujoy Basak, Arpit Babbar, Harish Kumar, Praveen Chandrashekar | Constraints Preserving Lax-Wendroff Flux Reconstruction for Relativistic Hydrodynamics with General Equations of State | null | null | null | null | math.NA cs.NA gr-qc | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | In the realm of relativistic astrophysics, the ideal equation of state with a constant adiabatic index provides a poor approximation due to its inconsistency with relativistic kinetic theory. However, it is a common practice to use it for relativistic fluid flow equations due to its simplicity. Here we develop a high-order Lax-Wendroff flux reconstruction method on Cartesian grids for solving special relativistic hydrodynamics equations with several general equations of state available in the literature. We also study the conversion from conservative to primitive variables, which depends on the equation of state in use, and provide an alternative method of conversion when the existing approach does not succeed. For the admissibility of the solution, we blend the high-order method with a low-order method on sub-cells and prove its physical admissible property in the case of all the equations of state used here. Lastly, we validate the scheme by several test cases having strong discontinuities, large Lorentz factor, and low density or pressure in one and two dimensions. | [
{
"version": "v1",
"created": "Thu, 08 May 2025 11:00:34 GMT"
},
{
"version": "v2",
"created": "Thu, 29 May 2025 14:33:29 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Basak",
"Sujoy",
""
],
[
"Babbar",
"Arpit",
""
],
[
"Kumar",
"Harish",
""
],
[
"Chandrashekar",
"Praveen",
""
]
] |
2505.05386 | Olivier Maillet | Aifei Zhang, Kenji Watanabe, Takashi Taniguchi, Patrice Roche, Carles Altimiras, Fran\c{c}ois D. Parmentier, Olivier Maillet | Ballistic-to-diffusive transition in engineered counter-propagating quantum Hall channels | 19 pages (6 main + 13 SI), 11 figures (4 main + 7 SI). SI + a few references added in v2 | null | null | null | cond-mat.mes-hall | http://creativecommons.org/licenses/by/4.0/ | Exotic quantum Hall systems hosting counter-propagating edge states can show seemingly non-universal transport regimes, usually depending on the size of the sample. We experimentally probe transport in a quantum Hall sample engineered to host a tunable number of counter-propagating edge states. The latter are coupled by Landauer reservoirs, which force charge equilibration over a tunable effective length. We show that charge transport is determined by the balance of up- and downstream channels, with a ballistic regime emerging for unequal numbers of channels. For equal numbers, we observe a transition to a critical diffusive regime, characterized by a diverging equilibration length. Our approach allows simulating the equilibration of hole-conjugate states and other exotic quantum Hall effects with fully controlled parameters using well-understood quantum Hall states. | [
{
"version": "v1",
"created": "Thu, 08 May 2025 16:19:55 GMT"
},
{
"version": "v2",
"created": "Wed, 28 May 2025 21:34:42 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Zhang",
"Aifei",
""
],
[
"Watanabe",
"Kenji",
""
],
[
"Taniguchi",
"Takashi",
""
],
[
"Roche",
"Patrice",
""
],
[
"Altimiras",
"Carles",
""
],
[
"Parmentier",
"François D.",
""
],
[
"Maillet",
"Olivier",
""
]
] |
2505.06186 | Massimiliano Pronesti | Massimiliano Pronesti, Joao Bettencourt-Silva, Paul Flanagan, Alessandra Pascale, Oisin Redmond, Anya Belz, Yufang Hou | Query-driven Document-level Scientific Evidence Extraction from Biomedical Studies | null | null | null | null | cs.CL cs.AI | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Extracting scientific evidence from biomedical studies for clinical research questions (e.g., Does stem cell transplantation improve quality of life in patients with medically refractory Crohn's disease compared to placebo?) is a crucial step in synthesising biomedical evidence. In this paper, we focus on the task of document-level scientific evidence extraction for clinical questions with conflicting evidence. To support this task, we create a dataset called CochraneForest, leveraging forest plots from Cochrane systematic reviews. It comprises 202 annotated forest plots, associated clinical research questions, full texts of studies, and study-specific conclusions. Building on CochraneForest, we propose URCA (Uniform Retrieval Clustered Augmentation), a retrieval-augmented generation framework designed to tackle the unique challenges of evidence extraction. Our experiments show that URCA outperforms the best existing methods by up to 10.3% in F1 score on this task. However, the results also underscore the complexity of CochraneForest, establishing it as a challenging testbed for advancing automated evidence synthesis systems. | [
{
"version": "v1",
"created": "Fri, 09 May 2025 16:55:06 GMT"
},
{
"version": "v2",
"created": "Tue, 13 May 2025 10:50:45 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Pronesti",
"Massimiliano",
""
],
[
"Bettencourt-Silva",
"Joao",
""
],
[
"Flanagan",
"Paul",
""
],
[
"Pascale",
"Alessandra",
""
],
[
"Redmond",
"Oisin",
""
],
[
"Belz",
"Anya",
""
],
[
"Hou",
"Yufang",
""
]
] |
2505.06612 | Yuqin Lan | Yuqin Lan and Laurence T. Yang | Burger: Robust Graph Denoising-augmentation Fusion and Multi-semantic Modeling in Social Recommendation | 10 pages, 5 figures | null | null | null | cs.SI cs.AI cs.IR | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | In the era of rapid development of social media, social recommendation systems as hybrid recommendation systems have been widely applied. Existing methods capture interest similarity between users to filter out interest-irrelevant relations in social networks that inevitably decrease recommendation accuracy, however, limited research has a focus on the mutual influence of semantic information between the social network and the user-item interaction network for further improving social recommendation. To address these issues, we introduce a social \underline{r}ecommendation model with ro\underline{bu}st g\underline{r}aph denoisin\underline{g}-augmentation fusion and multi-s\underline{e}mantic Modeling(Burger). Specifically, we firstly propose to construct a social tensor in order to smooth the training process of the model. Then, a graph convolutional network and a tensor convolutional network are employed to capture user's item preference and social preference, respectively. Considering the different semantic information in the user-item interaction network and the social network, a bi-semantic coordination loss is proposed to model the mutual influence of semantic information. To alleviate the interference of interest-irrelevant relations on multi-semantic modeling, we further use Bayesian posterior probability to mine potential social relations to replace social noise. Finally, the sliding window mechanism is utilized to update the social tensor as the input for the next iteration. Extensive experiments on three real datasets show Burger has a superior performance compared with the state-of-the-art models. | [
{
"version": "v1",
"created": "Sat, 10 May 2025 11:51:22 GMT"
},
{
"version": "v2",
"created": "Thu, 29 May 2025 16:52:21 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Lan",
"Yuqin",
""
],
[
"Yang",
"Laurence T.",
""
]
] |
2505.06750 | Parasara Sridhar Duggirala | Parasara Sridhar Duggirala and P. S. Thiagarajan | LPrL: An Asynchronous Linear Time Hyper Logic | null | null | null | null | cs.LO | http://creativecommons.org/licenses/by-nc-nd/4.0/ | We present a novel asynchronous hyper linear time temporal logic named LPrL (Linear Time Predicate Logic) and establish its basic theory. LPrL is a natural first order extension of LTL (Linear time temporal logic), in which the predicates specify the properties of and the relationships between traces (infinite sequences of actions) using Boolean combinations of LTL formulas. To augment the expressive power of the logic, we introduce a simple language of terms and add the equality predicate t = t' where t and t' are terms. We first illustrate how a number of the security policies as well as a basic consistency property of distributed processes can be captured using LPrL. We then establish our main results using automata theoretic techniques. Namely, the satisfiability and model checking problems for LPrL can be solved in elementary time. These results are in sharp contrast to HyperLTL, the prevalent synchronous hyper linear time logic, whose satisfiability problem is undecidable and whose model checking problem has non-elementary time complexity. | [
{
"version": "v1",
"created": "Sat, 10 May 2025 20:36:44 GMT"
},
{
"version": "v2",
"created": "Thu, 29 May 2025 13:23:27 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Duggirala",
"Parasara Sridhar",
""
],
[
"Thiagarajan",
"P. S.",
""
]
] |
2505.07655 | Joan Elias Mir\'o | Joan Elias Miro and Olivier Delouche | Flowing from the Ising Model on the Fuzzy Sphere to the 3D Lee-Yang CFT | added further details on the numerical data. logic unchanged | null | null | null | hep-th cond-mat.stat-mech cond-mat.str-el hep-lat | http://creativecommons.org/licenses/by/4.0/ | We employ the Fuzzy Sphere regulator to study the 3D Lee-Yang CFT. The model is defined by deforming the Ising model on the Fuzzy Sphere via a purely imaginary longitudinal magnetic field. This model undergoes a quantum phase transition, whose critical point we determine and identify with the 3D Lee-Yang CFT. We show how to tune the model and find that the lowest-lying states of the Hamiltonian align well with the expected CFT spectrum. We discuss the Fuzzy Sphere estimates for the scaling dimension $\Delta_\phi$ of the lowest primary operator. Finally, we interpret small deviations from the CFT expectations in terms of the leading irrelevant operators of the Lee-Yang CFT. We show that the Fuzzy Sphere calculations are compatible with the best five-loop $\epsilon$-expansion estimates. | [
{
"version": "v1",
"created": "Mon, 12 May 2025 15:22:56 GMT"
},
{
"version": "v2",
"created": "Thu, 29 May 2025 15:30:35 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Miro",
"Joan Elias",
""
],
[
"Delouche",
"Olivier",
""
]
] |
2505.07683 | Steven Song | Steven Song, Morgan Borjigin-Wang, Irene Madejski, Robert L. Grossman | Multimodal Survival Modeling in the Age of Foundation Models | 23 pages, 7 figures, 8 tables, updated with acknowledgements and declarations of interest | null | null | null | cs.LG cs.AI | http://creativecommons.org/licenses/by/4.0/ | The Cancer Genome Atlas (TCGA) has enabled novel discoveries and served as a large-scale reference through its harmonized genomics, clinical, and image data. Prior studies have trained bespoke cancer survival prediction models from unimodal or multimodal TCGA data. A modern paradigm in biomedical deep learning is the development of foundation models (FMs) to derive meaningful feature embeddings, agnostic to a specific modeling task. Biomedical text especially has seen growing development of FMs. While TCGA contains free-text data as pathology reports, these have been historically underutilized. Here, we investigate the feasibility of training classical, multimodal survival models over zero-shot embeddings extracted by FMs. We show the ease and additive effect of multimodal fusion, outperforming unimodal models. We demonstrate the benefit of including pathology report text and rigorously evaluate the effect of model-based text summarization and hallucination. Overall, we modernize survival modeling by leveraging FMs and information extraction from pathology reports. | [
{
"version": "v1",
"created": "Mon, 12 May 2025 15:47:21 GMT"
},
{
"version": "v2",
"created": "Wed, 28 May 2025 21:05:07 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Song",
"Steven",
""
],
[
"Borjigin-Wang",
"Morgan",
""
],
[
"Madejski",
"Irene",
""
],
[
"Grossman",
"Robert L.",
""
]
] |
2505.07889 | Yuyang Liu | Yuyang Liu, Liuzhenghao Lv, Xiancheng Zhang, Li Yuan, Yonghong Tian | BioProBench: Comprehensive Dataset and Benchmark in Biological Protocol Understanding and Reasoning | null | null | null | null | cs.CL | http://creativecommons.org/licenses/by-nc-nd/4.0/ | Biological protocols are fundamental to reproducibility and safety in life science research. While large language models (LLMs) perform well on general tasks, their systematic evaluation on these highly specialized, accuracy-critical, and inherently procedural texts remains limited. In this work, we present BioProBench, the first large-scale, multi-task benchmark for biological protocol understanding and reasoning. While there are several benchmark tasks involving protocol question answering, BioProBench provides a comprehensive suite of five core tasks: Protocol Question Answering, Step Ordering, Error Correction, Protocol Generation, and Protocol Reasoning, enabling a holistic evaluation of LLMs on procedural biological texts. Built upon 27K original protocols, it yields nearly 556K high-quality structured instances. We evaluate 12 mainstream open/closed-source LLMs. Experimental results reveal that some models perform well on basic understanding tasks (e.g., \sim70% PQA-Acc., >64% ERR F1), but struggle significantly with deep reasoning and structured generation tasks like ordering and generation. Furthermore, model comparisons show diverse performance: certain open-source models approach closed-source levels on some tasks, yet bio-specific small models lag behind general LLMs, indicating limitations on complex procedural content. Overall, BioProBench, through its task design and experimental findings, systematically reveals the fundamental challenges for current LLMs in procedural knowledge understanding, deep adaptability to specific domains, reliability of structured reasoning, and handling of sophisticated precision and safety constraints, providing key directions for future AI in the field of scientific experiment automation. The code and data are available at: https://github.com/YuyangSunshine/bioprotocolbench and https://huggingface.co/datasets/BioProBench/BioProBench. | [
{
"version": "v1",
"created": "Sun, 11 May 2025 09:42:24 GMT"
},
{
"version": "v2",
"created": "Thu, 29 May 2025 07:31:28 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Liu",
"Yuyang",
""
],
[
"Lv",
"Liuzhenghao",
""
],
[
"Zhang",
"Xiancheng",
""
],
[
"Yuan",
"Li",
""
],
[
"Tian",
"Yonghong",
""
]
] |
2505.08167 | Ruilin Liu | Ruilin Liu, Zhixiao Zhao, Jieqiong Li, Chang Liu, Dongbo Wang | Fusing Bidirectional Chains of Thought and Reward Mechanisms A Method for Enhancing Question-Answering Capabilities of Large Language Models for Chinese Intangible Cultural Heritage | 22 pages, 5 figures | null | null | null | cs.CL cs.AI | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | The rapid development of large language models (LLMs) has provided significant support and opportunities for the advancement of domain-specific LLMs. However, fine-tuning these large models using Intangible Cultural Heritage (ICH) data inevitably faces challenges such as bias, incorrect knowledge inheritance, and catastrophic forgetting. To address these issues, we propose a novel training method that integrates a bidirectional chains of thought and a reward mechanism. This method is built upon ICH-Qwen, a large language model specifically designed for the field of intangible cultural heritage. The proposed method enables the model to not only perform forward reasoning but also enhances the accuracy of the generated answers by utilizing reverse questioning and reverse reasoning to activate the model's latent knowledge. Additionally, a reward mechanism is introduced during training to optimize the decision-making process. This mechanism improves the quality of the model's outputs through structural and content evaluations with different weighting schemes. We conduct comparative experiments on ICH-Qwen, with results demonstrating that our method outperforms 0-shot, step-by-step reasoning, knowledge distillation, and question augmentation methods in terms of accuracy, Bleu-4, and Rouge-L scores on the question-answering task. Furthermore, the paper highlights the effectiveness of combining the bidirectional chains of thought and reward mechanism through ablation experiments. In addition, a series of generalizability experiments are conducted, with results showing that the proposed method yields improvements on various domain-specific datasets and advanced models in areas such as Finance, Wikidata, and StrategyQA. This demonstrates that the method is adaptable to multiple domains and provides a valuable approach for model training in future applications across diverse fields. | [
{
"version": "v1",
"created": "Tue, 13 May 2025 02:05:25 GMT"
},
{
"version": "v2",
"created": "Wed, 14 May 2025 01:35:33 GMT"
},
{
"version": "v3",
"created": "Thu, 29 May 2025 08:27:23 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Liu",
"Ruilin",
""
],
[
"Zhao",
"Zhixiao",
""
],
[
"Li",
"Jieqiong",
""
],
[
"Liu",
"Chang",
""
],
[
"Wang",
"Dongbo",
""
]
] |
2505.08249 | Mario Navarro Rodriguez | Mario Navarro-Rodriguez, Paul Philip Schmidt, Regina Hoffmann-Vogel, Andres M. Somoza and Elisa Palacios-Lidon | Modelling of time-dependent electrostatic effects and AFM-based surface conductivity characterization | null | null | null | null | cond-mat.mes-hall cond-mat.mtrl-sci physics.app-ph | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Atomic Force Microscopy (AFM) combined with electrical modes provides a powerful contactless approach to characterize material electrical properties at the nanoscale. However, conventional electrostatic models often overlook dynamic charge effects, which are particularly relevant for 2D materials deposited on insulating substrates. In this work, we introduce a theoretical framework that extends traditional electrostatic models by incorporating charge dynamics, analyzing two key cases: quasi-ideal conductors and quasi-ideal insulators. Our model establishes a characteristic timescale, $\tau$, which governs charge redistribution and measurement reliability. Experimental validation using Graphene Oxide, Reduced Graphene Oxide, and lightly reduced GO demonstrates strong dependence of frequency shift on surface conductivity, confirming our predictions. Temperature-dependent measurements further reveal conductivity variations consistent with disordered electronic materials. These findings provide critical insight into the impact of finite surface conductivity on AFM-based techniques and establish a novel method for evaluating charge dynamics in individual flakes of 2D materials and propose an alternative, contactless method for estimating surface conductivity. | [
{
"version": "v1",
"created": "Tue, 13 May 2025 05:57:24 GMT"
},
{
"version": "v2",
"created": "Thu, 29 May 2025 12:44:37 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Navarro-Rodriguez",
"Mario",
""
],
[
"Schmidt",
"Paul Philip",
""
],
[
"Hoffmann-Vogel",
"Regina",
""
],
[
"Somoza",
"Andres M.",
""
],
[
"Palacios-Lidon",
"Elisa",
""
]
] |
2505.08408 | Guangxuan Lin | Guangxuan Lin and Shouqiang Du | A three-term Polak-Ribi\`{e}re-Polyak conjugate gradient method for vector optimization | null | null | null | null | math.OC | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | A three-term Polak-Ribi\`{e}re-Polyak conjugate gradient method is first proposed in this paper for solving vector optimization problems. This method can autonomously generate descent directions independent of line search procedures, while retaining the conjugate property. The global convergence of the proposed scheme is established by the generalized Wolfe line search procedure without self-adjusting strategies, regular restarts and convexity assumptions. Finally, numerical experiments illustrating the practical performance of this method are presented. | [
{
"version": "v1",
"created": "Tue, 13 May 2025 10:08:12 GMT"
},
{
"version": "v2",
"created": "Thu, 29 May 2025 09:26:14 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Lin",
"Guangxuan",
""
],
[
"Du",
"Shouqiang",
""
]
] |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.