# Research Papers — 2025-12 · bundle page 29 of 175 > Full-text markdown of papers #1401–1450 of 8703 published in 2025-12, newest first. AI-readable mirrors of the full 8703-paper 2025-12 archive (50 papers per file). Canonical page: https://aipod.dev/papers/archive/2025-12/page/29 [← Previous bundle](https://aipod.dev/papers/archive/2025-12/page/28.md) [Next bundle →](https://aipod.dev/papers/archive/2025-12/page/30.md) ### Fast SAM2 with Text-Driven Token Pruning Authors: Avilasha Mandal, Chaoning Zhang, Fachrina Dewi Puspitasari, Xudong Wang, Jiaquan Zhang, Caiyan Qin, Guoqing Wang, Yang Yang, Heng Tao Shen Published: 2025-12-24 Categories: cs.CV arXiv: https://arxiv.org/abs/2512.21333v1 Segment Anything Model 2 (SAM2), a vision foundation model has significantly advanced in prompt-driven video object segmentation, yet their practical deployment remains limited by the high computational and memory cost of processing dense visual tokens across time. The SAM2 pipelines typically propagate all visual tokens produced by the image encoder through downstream temporal reasoning modules, regardless of their relevance to the target object, resulting in reduced scalability due to quadratic memory attention overhead. In this work, we introduce a text-guided token pruning framework that improves inference efficiency by selectively reducing token density prior to temporal propagation, without modifying the underlying segmentation architecture. Operating after visual encoding and before memory based propagation, our method ranks tokens using a lightweight routing mechanism that integrates local visual context, semantic relevance derived from object-centric textual descriptions (either user-provided or automatically generated), and uncertainty cues that help preserve ambiguous or boundary critical regions. By retaining only the most informative tokens for downstream processing, the proposed approach reduces redundant computation while maintaining segmentation fidelity. Extensive experiments across multiple challenging video segmentation benchmarks demonstrate that post-encoder token pruning provides a practical and effective pathway to efficient, prompt-aware video segmentation, achieving up to 42.50 percent faster inference and 37.41 percent lower GPU memory usage compared to the unpruned baseline SAM2, while preserving competitive J and F performance. These results highlight the potential of early token selection to improve the scalability of transformer-based video segmentation systems for real-time and resource-constrained applications. --- ### C2LLM Technical Report: A New Frontier in Code Retrieval via Adaptive Cross-Attention Pooling Authors: Jin Qin, Zihan Liao, Ziyin Zhang, Hang Yu, Peng Di, Rui Wang Published: 2025-12-24 Categories: cs.CL, cs.AI arXiv: https://arxiv.org/abs/2512.21332v1 We present C2LLM - Contrastive Code Large Language Models, a family of code embedding models in both 0.5B and 7B sizes. Building upon Qwen-2.5-Coder backbones, C2LLM adopts a Pooling by Multihead Attention (PMA) module for generating sequence embedding from token embeddings, effectively 1) utilizing the LLM's causal representations acquired during pretraining, while also 2) being able to aggregate information from all tokens in the sequence, breaking the information bottleneck in EOS-based sequence embeddings, and 3) supporting flexible adaptation of embedding dimension, serving as an alternative to MRL. Trained on three million publicly available data, C2LLM models set new records on MTEB-Code among models of similar sizes, with C2LLM-7B ranking 1st on the overall leaderboard. --- ### TICON: A Slide-Level Tile Contextualizer for Histopathology Representation Learning Authors: Varun Belagali, Saarthak Kapse, Pierre Marza, Srijan Das, Zilinghan Li, Sofiène Boutaj, Pushpak Pati, Srikar Yellapragada, Tarak Nath Nandi, Ravi K Madduri, Joel Saltz, Prateek Prasanna et al. Published: 2025-12-24 Categories: cs.CV arXiv: https://arxiv.org/abs/2512.21331v2 The interpretation of small tiles in large whole slide images (WSI) often needs a larger image context. We introduce TICON, a transformer-based tile representation contextualizer that produces rich, contextualized embeddings for ''any'' application in computational pathology. Standard tile encoder-based pipelines, which extract embeddings of tiles stripped from their context, fail to model the rich slide-level information essential for both local and global tasks. Furthermore, different tile-encoders excel at different downstream tasks. Therefore, a unified model is needed to contextualize embeddings derived from ''any'' tile-level foundation model. TICON addresses this need with a single, shared encoder, pretrained using a masked modeling objective to simultaneously unify and contextualize representations from diverse tile-level pathology foundation models. Our experiments demonstrate that TICON-contextualized embeddings significantly improve performance across many different tasks, establishing new state-of-the-art results on tile-level benchmarks (i.e., HEST-Bench, THUNDER, CATCH) and slide-level benchmarks (i.e., Patho-Bench). Finally, we pretrain an aggregator on TICON to form a slide-level foundation model, using only 11K WSIs, outperforming SoTA slide-level foundation models pretrained with up to 350K WSIs. --- ### TICON: A Slide-Level Tile Contextualizer for Histopathology Representation Learning Authors: Varun Belagali, Saarthak Kapse, Pierre Marza, Srijan Das, Zilinghan Li, Sofiène Boutaj, Pushpak Pati, Srikar Yellapragada, Tarak Nath Nandi, Ravi K Madduri, Joel Saltz, Prateek Prasanna et al. Published: 2025-12-24 Categories: cs.CV arXiv: https://arxiv.org/abs/2512.21331v1 The interpretation of small tiles in large whole slide images (WSI) often needs a larger image context. We introduce TICON, a transformer-based tile representation contextualizer that produces rich, contextualized embeddings for ''any'' application in computational pathology. Standard tile encoder-based pipelines, which extract embeddings of tiles stripped from their context, fail to model the rich slide-level information essential for both local and global tasks. Furthermore, different tile-encoders excel at different downstream tasks. Therefore, a unified model is needed to contextualize embeddings derived from ''any'' tile-level foundation model. TICON addresses this need with a single, shared encoder, pretrained using a masked modeling objective to simultaneously unify and contextualize representations from diverse tile-level pathology foundation models. Our experiments demonstrate that TICON-contextualized embeddings significantly improve performance across many different tasks, establishing new state-of-the-art results on tile-level benchmarks (i.e., HEST-Bench, THUNDER, CATCH) and slide-level benchmarks (i.e., Patho-Bench). Finally, we pretrain an aggregator on TICON to form a slide-level foundation model, using only 11K WSIs, outperforming SoTA slide-level foundation models pretrained with up to 350K WSIs. --- ### Your Reasoning Benchmark May Not Test Reasoning: Revealing Perception Bottleneck in Abstract Reasoning Benchmarks Authors: Xinhe Wang, Jin Huang, Xingjian Zhang, Tianhao Wang, Jiaqi W. Ma Published: 2025-12-24 Categories: cs.CL arXiv: https://arxiv.org/abs/2512.21329v2 Reasoning benchmarks such as the Abstraction and Reasoning Corpus (ARC) and ARC-AGI are widely used to assess progress in artificial intelligence and are often interpreted as probes of core, so-called ``fluid'' reasoning abilities. Despite their apparent simplicity for humans, these tasks remain challenging for frontier vision-language models (VLMs), a gap commonly attributed to deficiencies in machine reasoning. We challenge this interpretation and hypothesize that the gap arises primarily from limitations in visual perception rather than from shortcomings in inductive reasoning. To verify this hypothesis, we introduce a two-stage experimental pipeline that explicitly separates perception and reasoning. In the perception stage, each image is independently converted into a natural-language description, while in the reasoning stage a model induces and applies rules using these descriptions. This design prevents leakage of cross-image inductive signals and isolates reasoning from perception bottlenecks. Across three ARC-style datasets, Mini-ARC, ACRE, and Bongard-LOGO, we show that the perception capability is the dominant factor underlying the observed performance gap by comparing the two-stage pipeline with against standard end-to-end one-stage evaluation. Manual inspection of reasoning traces in the VLM outputs further reveals that approximately 80 percent of model failures stem from perception errors. Together, these results demonstrate that ARC-style benchmarks conflate perceptual and reasoning challenges and that observed performance gaps may overstate deficiencies in machine reasoning. Our findings underscore the need for evaluation protocols that disentangle perception from reasoning when assessing progress in machine intelligence. --- ### Your Reasoning Benchmark May Not Test Reasoning: Revealing Perception Bottleneck in Abstract Reasoning Benchmarks Authors: Xinhe Wang, Jin Huang, Xingjian Zhang, Tianhao Wang, Jiaqi W. Ma Published: 2025-12-24 Categories: cs.CL arXiv: https://arxiv.org/abs/2512.21329v1 Reasoning benchmarks such as the Abstraction and Reasoning Corpus (ARC) and ARC-AGI are widely used to assess progress in artificial intelligence and are often interpreted as probes of core, so-called ``fluid'' reasoning abilities. Despite their apparent simplicity for humans, these tasks remain challenging for frontier vision-language models (VLMs), a gap commonly attributed to deficiencies in machine reasoning. We challenge this interpretation and hypothesize that the gap arises primarily from limitations in visual perception rather than from shortcomings in inductive reasoning. To verify this hypothesis, we introduce a two-stage experimental pipeline that explicitly separates perception and reasoning. In the perception stage, each image is independently converted into a natural-language description, while in the reasoning stage a model induces and applies rules using these descriptions. This design prevents leakage of cross-image inductive signals and isolates reasoning from perception bottlenecks. Across three ARC-style datasets, Mini-ARC, ACRE, and Bongard-LOGO, we show that the perception capability is the dominant factor underlying the observed performance gap by comparing the two-stage pipeline with against standard end-to-end one-stage evaluation. Manual inspection of reasoning traces in the VLM outputs further reveals that approximately 80 percent of model failures stem from perception errors. Together, these results demonstrate that ARC-style benchmarks conflate perceptual and reasoning challenges and that observed performance gaps may overstate deficiencies in machine reasoning. Our findings underscore the need for evaluation protocols that disentangle perception from reasoning when assessing progress in machine intelligence. --- ### Measuring all the noises of LLM Evals Authors: Sida I. Wang Published: 2025-12-24 Categories: cs.LG, cs.AI, cs.CL, stat.ML arXiv: https://arxiv.org/abs/2512.21326v3 Applying well-established statistical methods effectively to LLM evals requires consideration of model correlations and prediction noises. Leading evals often ignore data noise, have low statistical power, or make unnecessary trade-offs. We define and measure three types of noise: 1) prediction noise from generating different answers on a given question, 2) data noise from sampling questions, and 3) their combined total noise following the law of total variance. To emphasize relative comparisons and gain statistical power, we propose \emph{all-pairs comparison}, which applies paired comparison to all pairs of LLMs and measures all the noise components based on millions of question-level predictions across many evals and settings, revealing clear trends. First, most evals exhibit a characteristic and highly predictable total noise level across all model pairs. Second, paired prediction noise typically exceeds paired data noise, which means reducing prediction noise by averaging can significantly increase statistical power. By measuring all the noises together, we can assess eval results in context and make better empirical decisions. --- ### Measuring all the noises of LLM Evals Authors: Sida Wang Published: 2025-12-24 Categories: cs.LG, cs.AI, cs.CL, stat.ML arXiv: https://arxiv.org/abs/2512.21326v1 Separating signal from noise is central to experimental science. Applying well-established statistical method effectively to LLM evals requires consideration of their unique noise characteristics. We clearly define and measure three types of noise: prediction noise from generating different answers on a given question, data noise from sampling questions, and their combined total noise following the law of total variance. To emphasize relative comparisons and gain statistical power, we propose the all-pairs paired method, which applies the paired analysis to all pairs of LLMs and measures all the noise components based on millions of question-level predictions across many evals and settings. These measurements revealed clear patterns. First, each eval exhibits a characteristic and highly predictable total noise level across all model pairs. Second, paired prediction noise typically exceeds paired data noise, which means reducing prediction noise by averaging can significantly increase statistical power. These findings enable practitioners to assess significance without custom testing and to detect much smaller effects in controlled experiments. --- ### Parallel Token Prediction for Language Models Authors: Felix Draxler, Justus Will, Farrin Marouf Sofian, Theofanis Karaletsos, Sameer Singh, Stephan Mandt Published: 2025-12-24 Categories: cs.CL, cs.LG arXiv: https://arxiv.org/abs/2512.21323v2 Autoregressive decoding in language models is inherently slow, generating only one token per forward pass. We propose Parallel Token Prediction (PTP), a general-purpose framework for predicting multiple tokens in a single model call. PTP moves the source of randomness from post-hoc sampling to random input variables, making future tokens deterministic functions of those inputs and thus jointly predictable in a single forward pass. We prove that a single PTP call can represent arbitrary dependencies between tokens. PTP is trained by distilling an existing model or through inverse autoregressive training without a teacher. Experimentally, PTP achieves a 2.4x speedup on a diverse-task speculative decoding benchmark. We provide code and checkpoints at https://github.com/mandt-lab/ptp. --- ### Parallel Token Prediction for Language Models Authors: Felix Draxler, Justus Will, Farrin Marouf Sofian, Theofanis Karaletsos, Sameer Singh, Stephan Mandt Published: 2025-12-24 Categories: cs.CL, cs.LG arXiv: https://arxiv.org/abs/2512.21323v1 We propose Parallel Token Prediction (PTP), a universal framework for parallel sequence generation in language models. PTP jointly predicts multiple dependent tokens in a single transformer call by incorporating the sampling procedure into the model. This reduces the latency bottleneck of autoregressive decoding, and avoids the restrictive independence assumptions common in existing multi-token prediction methods. We prove that PTP can represent arbitrary autoregressive sequence distributions. PTP is trained either by distilling an existing model or through inverse autoregressive training without a teacher. Experimentally, we achieve state-of-the-art speculative decoding performance on Vicuna-7B by accepting over four tokens per step on Spec-Bench. The universality of our framework indicates that parallel generation of long sequences is feasible without loss of modeling power. --- ### Variationally correct operator learning: Reduced basis neural operator with a posteriori error estimation Authors: Yuan Qiu, Wolfgang Dahmen, Peng Chen Published: 2025-12-24 Categories: math.NA, cs.LG arXiv: https://arxiv.org/abs/2512.21319v1 Minimizing PDE-residual losses is a common strategy to promote physical consistency in neural operators. However, standard formulations often lack variational correctness, meaning that small residuals do not guarantee small solution errors due to the use of non-compliant norms or ad hoc penalty terms for boundary conditions. This work develops a variationally correct operator learning framework by constructing first-order system least-squares (FOSLS) objectives whose values are provably equivalent to the solution error in PDE-induced norms. We demonstrate this framework on stationary diffusion and linear elasticity, incorporating mixed Dirichlet-Neumann boundary conditions via variational lifts to preserve norm equivalence without inconsistent penalties. To ensure the function space conformity required by the FOSLS loss, we propose a Reduced Basis Neural Operator (RBNO). The RBNO predicts coefficients for a pre-computed, conforming reduced basis, thereby ensuring variational stability by design while enabling efficient training. We provide a rigorous convergence analysis that bounds the total error by the sum of finite element discretization bias, reduced basis truncation error, neural network approximation error, and statistical estimation errors arising from finite sampling and optimization. Numerical benchmarks validate these theoretical bounds and demonstrate that the proposed approach achieves superior accuracy in PDE-compliant norms compared to standard baselines, while the residual loss serves as a reliable, computable a posteriori error estimator. --- ### Scaling Laws for Economic Productivity: Experimental Evidence in LLM-Assisted Consulting, Data Analyst, and Management Tasks Authors: Ali Merali Published: 2025-12-24 Categories: econ.GN, cs.AI, cs.HC arXiv: https://arxiv.org/abs/2512.21316v1 This paper derives `Scaling Laws for Economic Impacts' -- empirical relationships between the training compute of Large Language Models (LLMs) and professional productivity. In a preregistered experiment, over 500 consultants, data analysts, and managers completed professional tasks using one of 13 LLMs. We find that each year of AI model progress reduced task time by 8%, with 56% of gains driven by increased compute and 44% by algorithmic progress. However, productivity gains were significantly larger for non-agentic analytical tasks compared to agentic workflows requiring tool use. These findings suggest continued model scaling could boost U.S. productivity by approximately 20% over the next decade. --- ### Does the Data Processing Inequality Reflect Practice? On the Utility of Low-Level Tasks Authors: Roy Turgeman, Tom Tirer Published: 2025-12-24 Categories: cs.LG, cs.CV, stat.ML arXiv: https://arxiv.org/abs/2512.21315v2 The data processing inequality is an information-theoretic principle stating that the information content of a signal cannot be increased by processing the observations. In particular, it suggests that there is no benefit in enhancing the signal or encoding it before addressing a classification problem. This assertion can be proven to be true for the case of the optimal Bayes classifier. However, in practice, it is common to perform "low-level" tasks before "high-level" downstream tasks despite the overwhelming capabilities of modern deep neural networks. In this paper, we aim to understand when and why low-level processing can be beneficial for classification. We present a comprehensive theoretical study of a binary classification setup, where we consider a classifier that is tightly connected to the optimal Bayes classifier and converges to it as the number of training samples increases. We prove that for any finite number of training samples, there exists a pre-classification processing that improves the classification accuracy. We also explore the effect of class separation, training set size, and class balance on the relative gain from this procedure. We support our theory with an empirical investigation of the theoretical setup. Finally, we conduct an empirical study where we investigate the effect of denoising and encoding on the performance of practical deep classifiers on benchmark datasets. Specifically, we vary the size and class distribution of the training set, and the noise level, and demonstrate trends that are consistent with our theoretical results. --- ### Does the Data Processing Inequality Reflect Practice? On the Utility of Low-Level Tasks Authors: Roy Turgeman, Tom Tirer Published: 2025-12-24 Categories: cs.LG, cs.CV, stat.ML arXiv: https://arxiv.org/abs/2512.21315v1 The data processing inequality is an information-theoretic principle stating that the information content of a signal cannot be increased by processing the observations. In particular, it suggests that there is no benefit in enhancing the signal or encoding it before addressing a classification problem. This assertion can be proven to be true for the case of the optimal Bayes classifier. However, in practice, it is common to perform "low-level" tasks before "high-level" downstream tasks despite the overwhelming capabilities of modern deep neural networks. In this paper, we aim to understand when and why low-level processing can be beneficial for classification. We present a comprehensive theoretical study of a binary classification setup, where we consider a classifier that is tightly connected to the optimal Bayes classifier and converges to it as the number of training samples increases. We prove that for any finite number of training samples, there exists a pre-classification processing that improves the classification accuracy. We also explore the effect of class separation, training set size, and class balance on the relative gain from this procedure. We support our theory with an empirical investigation of the theoretical setup. Finally, we conduct an empirical study where we investigate the effect of denoising and encoding on the performance of practical deep classifiers on benchmark datasets. Specifically, we vary the size and class distribution of the training set, and the noise level, and demonstrate trends that are consistent with our theoretical results. --- ### Learning to Solve PDEs on Neural Shape Representations Authors: Lilian Welschinger, Yilin Liu, Zican Wang, Niloy Mitra Published: 2025-12-24 Categories: cs.LG arXiv: https://arxiv.org/abs/2512.21311v2 Solving partial differential equations (PDEs) on shapes underpins many shape analysis and engineering tasks; yet, prevailing PDE solvers operate on polygonal/triangle meshes while modern 3D assets increasingly live as neural representations. This mismatch leaves no suitable method to solve surface PDEs directly within the neural domain, forcing explicit mesh extraction or per-instance residual training, preventing end-to-end workflows. We present a novel, meshfree formulation that learns a local update operator conditioned on neural (local) shape attributes, enabling surface PDEs to be solved directly where the (neural) data lives. The operator integrates naturally with prevalent neural surface representations, is trained once on a single representative shape, and generalizes across shape and topology variations, enabling accurate, fast inference without explicit meshing or per-instance optimization while preserving differentiability. Across analytic benchmarks (heat diffusion and Poisson equations on the sphere) and on diverse shapes and neural surface representations, our method achieves accuracy comparable to classical solvers while enabling a unified, end-to-end pipeline across neural and traditional surface representations. Our source code and project page: https://welschinger.github.io/Learning-to-Solve-PDEs-on-Neural-Shape-Representations/. --- ### Learning to Solve PDEs on Neural Shape Representations Authors: Lilian Welschinger, Yilin Liu, Zican Wang, Niloy Mitra Published: 2025-12-24 Categories: cs.LG arXiv: https://arxiv.org/abs/2512.21311v1 Solving partial differential equations (PDEs) on shapes underpins many shape analysis and engineering tasks; yet, prevailing PDE solvers operate on polygonal/triangle meshes while modern 3D assets increasingly live as neural representations. This mismatch leaves no suitable method to solve surface PDEs directly within the neural domain, forcing explicit mesh extraction or per-instance residual training, preventing end-to-end workflows. We present a novel, mesh-free formulation that learns a local update operator conditioned on neural (local) shape attributes, enabling surface PDEs to be solved directly where the (neural) data lives. The operator integrates naturally with prevalent neural surface representations, is trained once on a single representative shape, and generalizes across shape and topology variations, enabling accurate, fast inference without explicit meshing or per-instance optimization while preserving differentiability. Across analytic benchmarks (heat equation and Poisson solve on sphere) and real neural assets across different representations, our method slightly outperforms CPM while remaining reasonably close to FEM, and, to our knowledge, delivers the first end-to-end pipeline that solves surface PDEs on both neural and classical surface representations. Code will be released on acceptance. --- ### AndroidLens: Long-latency Evaluation with Nested Sub-targets for Android GUI Agents Authors: Yue Cao, Yingyao Wang, Pi Bu, Jingxuan Xing, Wei Jiang, Zekun Zhu, Junpeng Ma, Sashuai Zhou, Tong Lu, Jun Song, Yu Cheng, Yuning Jiang et al. Published: 2025-12-24 Categories: cs.CV arXiv: https://arxiv.org/abs/2512.21302v1 Graphical user interface (GUI) agents can substantially improve productivity by automating frequently executed long-latency tasks on mobile devices. However, existing evaluation benchmarks are still constrained to limited applications, simple tasks, and coarse-grained metrics. To address this, we introduce AndroidLens, a challenging evaluation framework for mobile GUI agents, comprising 571 long-latency tasks in both Chinese and English environments, each requiring an average of more than 26 steps to complete. The framework features: (1) tasks derived from real-world user scenarios across 38 domains, covering complex types such as multi-constraint, multi-goal, and domain-specific tasks; (2) static evaluation that preserves real-world anomalies and allows multiple valid paths to reduce bias; and (3) dynamic evaluation that employs a milestone-based scheme for fine-grained progress measurement via Average Task Progress (ATP). Our evaluation indicates that even the best models reach only a 12.7% task success rate and 50.47% ATP. We also underscore key challenges in real-world environments, including environmental anomalies, adaptive exploration, and long-term memory retention. --- ### Transcriptome-Conditioned Personalized De Novo Drug Generation for AML Using Metaheuristic Assembly and Target-Driven Filtering Authors: Abdullah G. Elafifi, Basma Mamdouh, Mariam Hanafy, Muhammed Alaa Eldin, Yosef Khaled, Nesma Mohamed El-Gelany, Tarek H. M. Abou-El-Enien Published: 2025-12-24 Categories: cs.LG, q-bio.QM arXiv: https://arxiv.org/abs/2512.21301v1 Acute Myeloid Leukemia (AML) remains a clinical challenge due to its extreme molecular heterogeneity and high relapse rates. While precision medicine has introduced mutation-specific therapies, many patients still lack effective, personalized options. This paper presents a novel, end-to-end computational framework that bridges the gap between patient-specific transcriptomics and de novo drug discovery. By analyzing bulk RNA sequencing data from the TCGA-LAML cohort, the study utilized Weighted Gene Co-expression Network Analysis (WGCNA) to prioritize 20 high-value biomarkers, including metabolic transporters like HK3 and immune-modulatory receptors such as SIGLEC9. The physical structures of these targets were modeled using AlphaFold3, and druggable hotspots were quantitatively mapped via the DOGSiteScorer engine. Then developed a novel, reaction-first evolutionary metaheuristic algorithm as well as multi-objective optimization programming that assembles novel ligands from fragment libraries, guided by spatial alignment to these identified hotspots. The generative model produced structurally unique chemical entities with a strong bias toward drug-like space, as evidenced by QED scores peaking between 0.5 and 0.7. Validation through ADMET profiling and SwissDock molecular docking identified high-confidence candidates, such as Ligand L1, which achieved a binding free energy of -6.571 kcal/mol against the A08A96 biomarker. These results demonstrate that integrating systems biology with metaheuristic molecular assembly can produce pharmacologically viable, patient tailored leads, offering a scalable blueprint for precision oncology in AML and beyond --- ### Model Merging via Multi-Teacher Knowledge Distillation Authors: Seyed Arshan Dalili, Mehrdad Mahdavi Published: 2025-12-24 Categories: cs.LG, cs.AI arXiv: https://arxiv.org/abs/2512.21288v1 Model merging has emerged as a lightweight alternative to joint multi-task learning (MTL), yet the generalization properties of merged models remain largely unexplored. Establishing such theoretical guarantees is non-trivial, as the merging process typically forbids access to the original training data and involves combining fine-tuned models trained on fundamentally heterogeneous data distributions. Without a principled understanding of these dynamics, current methods often rely on heuristics to approximate the optimal combination of parameters. This dependence is most critical in coefficient scaling, the weighting factors that modulate the magnitude of each fine-tuned model's contribution to the shared parameter. However, without a principled objective to guide their selection, these methods lead to brittle performance and are highly sensitive to scaling initialization. We address this gap by (i) establishing a novel flatness-aware PAC-Bayes generalization bound specifically for the model merging setting. This analysis introduces a "cross-task heterogeneity" term that formally captures the mismatch between diverse fine-tuned model priors and the target multi-task distributions. Guided by this theoretical insight, (ii) we frame model merging as multi-teacher knowledge distillation on scarce, unlabeled data. We formally demonstrate that minimizing the student-teacher Kullback-Leibler divergence directly tightens the upper bound on the merged model's excess risk. Guided by the flatness-aware bound derived, (iii) we operationalize this objective via SAMerging, a method that employs Sharpness-Aware Minimization (SAM) to find flat minima. Empirically, SAMerging establishes a new state of the art across vision and NLP benchmarks, achieving remarkable performance. The code is available at https://github.com/arshandalili/SAMerging. --- ### Post-Processing Mask-Based Table Segmentation for Structural Coordinate Extraction Authors: Suren Bandara Published: 2025-12-24 Categories: cs.CV arXiv: https://arxiv.org/abs/2512.21287v1 Structured data extraction from tables plays a crucial role in document image analysis for scanned documents and digital archives. Although many methods have been proposed to detect table structures and extract cell contents, accurately identifying table segment boundaries (rows and columns) remains challenging, particularly in low-resolution or noisy images. In many real-world scenarios, table data are incomplete or degraded, limiting the adaptability of transformer-based methods to noisy inputs. Mask-based edge detection techniques have shown greater robustness under such conditions, as their sensitivity can be adjusted through threshold tuning; however, existing approaches typically apply masks directly to images, leading to noise sensitivity, resolution loss, or high computational cost. This paper proposes a novel multi-scale signal-processing method for detecting table edges from table masks. Row and column transitions are modeled as one-dimensional signals and processed using Gaussian convolution with progressively increasing variances, followed by statistical thresholding to suppress noise while preserving stable structural edges. Detected signal peaks are mapped back to image coordinates to obtain accurate segment boundaries. Experimental results show that applying the proposed approach to column edge detection improves Cell-Aware Segmentation Accuracy (CASA) a layout-aware metric evaluating both textual correctness and correct cell placement from 67% to 76% on the PubLayNet-1M benchmark when using TableNet with PyTesseract OCR. The method is robust to resolution variations through zero-padding and scaling strategies and produces optimized structured tabular outputs suitable for downstream analysis. --- ### Toward Real-Time Surgical Scene Segmentation via a Spike-Driven Video Transformer with Spike-Informed Pretraining Authors: Shihao Zou, Jingjing Li, Wei Ji, Jincai Huang, Kai Wang, Guo Dan, Weixin Si, Yi Pan Published: 2025-12-24 Categories: cs.CV arXiv: https://arxiv.org/abs/2512.21284v2 Modern surgical systems increasingly rely on intelligent scene understanding to improve intra-operative safety and situational awareness, with surgical scene segmentation playing a fundamental role in fine-grained surgical perception. Although recent ANN models, especially large foundation models, have achieved impressive accuracy, their high computational and energy demands often hinder deployment in resource-constrained operative environments. To address this challenge, we explore SNN as a highly efficient paradigm. However, its performance in surgical scene segmentation remains constrained by sparse spike representations and limited annotated surgical data. We therefore propose SpikeSurgSeg, the first spike-driven video Transformer for surgical scene segmentation. It preserves the real-time and energy-efficient advantages of SNN, while achieving competitive performance against most ANN models in data-scarce surgical scenarios. Specifically, we introudce a spike-informed pretraining strategy based on MAE, where mask generation is guided by spike firing activity to better align with sparse spike representations, together with a layer-wise tube masking scheme that reduces information leakage and encourages contextual reasoning. To further strengthen semantic representation, we introduce multi-spectral knowledge distillation, which aligns teacher ANN and student SNN features in the frequency domain, where the mismatch between continuous activation patterns and spike-driven temporal representations can be effectively mitigated. Built on the pretrained SNN encoder, we further design a lightweight spike-driven segmentation head. Extensive experiments on EndoVis18 and our in-house SurgBleed dataset show that SpikeSurgSeg achieves mIoU comparable to SOTA ANN models while reducing inference latency by at least 8x. Notably, it delivers over 20x speedup relative to most foundation models. --- ### Surgical Scene Segmentation using a Spike-Driven Video Transformer with Real-Time Potential Authors: Shihao Zou, Jingjing Li, Wei Ji, Jincai Huang, Kai Wang, Guo Dan, Weixin Si, Yi Pan Published: 2025-12-24 Categories: cs.CV arXiv: https://arxiv.org/abs/2512.21284v1 Modern surgical systems increasingly rely on intelligent scene understanding to provide timely situational awareness for enhanced intra-operative safety. Within this pipeline, surgical scene segmentation plays a central role in accurately perceiving operative events. Although recent deep learning models, particularly large-scale foundation models, achieve remarkable segmentation accuracy, their substantial computational demands and power consumption hinder real-time deployment in resource-constrained surgical environments. To address this limitation, we explore the emerging SNN as a promising paradigm for highly efficient surgical intelligence. However, their performance is still constrained by the scarcity of labeled surgical data and the inherently sparse nature of surgical video representations. To this end, we propose \textit{SpikeSurgSeg}, the first spike-driven video Transformer framework tailored for surgical scene segmentation with real-time potential on non-GPU platforms. To address the limited availability of surgical annotations, we introduce a surgical-scene masked autoencoding pretraining strategy for SNNs that enables robust spatiotemporal representation learning via layer-wise tube masking. Building on this pretrained backbone, we further adopt a lightweight spike-driven segmentation head that produces temporally consistent predictions while preserving the low-latency characteristics of SNNs. Extensive experiments on EndoVis18 and our in-house SurgBleed dataset demonstrate that SpikeSurgSeg achieves mIoU comparable to SOTA ANN-based models while reducing inference latency by at least $8\times$. Notably, it delivers over $20\times$ acceleration relative to most foundation-model baselines, underscoring its potential for time-critical surgical scene segmentation. --- ### SMART SLM: Structured Memory and Reasoning Transformer, A Small Language Model for Accurate Document Assistance Authors: Divij Dudeja, Mayukha Pal Published: 2025-12-24 Categories: cs.CL, cs.AI arXiv: https://arxiv.org/abs/2512.21280v1 The user of Engineering Manuals (EM) finds it difficult to read EM s because they are long, have a dense format which includes written documents, step by step procedures, and standard parameter lists for engineering equipment. Off the shelf transformers, especially compact ones, treat this material as a flat stream of tokens. This approach leads to confident but incorrect numeric answers and forces the models to memorize separate facts inefficiently. SMART (Structured Memory and Reasoning Transformer) offers a different and practical solution to the above problem. SMART structures its processing by using a hierarchical approach, and is based upon three main job categories (1) A syntax-aware Fact Extractor (Grammarian) Tree LSTM which extracts facts as subject relation object relations from EM sentences (2) A compact indexed memory MANN (Memory Augmented Neural Network) that indexes these Rational Subject Relation Objects as 384 dimensional vectors that are associated with the source of the information, and (3) A 6 layer Transformer that learns to fuse the previously retrieved facts into its generated response. The entire SMART model utilizes 45.51M parameters, which is 64% less than GPT-2 (124M) and 69% less than BERT (133M), and it achieves a 21.3% higher accuracy than GPT-2, indicating that SMART fits the data better with the least amount of processing requirements. SMART employs dual modes of inference an indexed fast path for known documents (sub-second answer times) and an indexed dynamic path assisted by RAGs for new uploads (FAISS Top 20 results with memory severed at 64 slots). In real world deployment, this framework leads to more well supported results with reduced hallucinations than comparable small transformer models. --- ### GriDiT: Factorized Grid-Based Diffusion for Efficient Long Image Sequence Generation Authors: Snehal Singh Tomar, Alexandros Graikos, Arjun Krishna, Dimitris Samaras, Klaus Mueller Published: 2025-12-24 Categories: cs.CV arXiv: https://arxiv.org/abs/2512.21276v2 Modern deep learning methods typically treat image sequences as large tensors of sequentially stacked frames. However, is this straightforward representation ideal given the current state-of-the-art (SoTA)? In this work, we address this question in the context of generative models and aim to devise a more effective way of modeling image sequence data. Observing the inefficiencies and bottlenecks of current SoTA image sequence generation methods, we showcase that rather than working with large tensors, we can improve the generation process by factorizing it into first generating the coarse sequence at low resolution and then refining the individual frames at high resolution. We train a generative model solely on grid images comprising subsampled frames. Yet, we learn to generate image sequences, using the strong self-attention mechanism of the Diffusion Transformer (DiT) to capture correlations between frames. In effect, our formulation extends a 2D image generator to operate as a low-resolution 3D image-sequence generator without introducing any architectural modifications. Subsequently, we super-resolve each frame individually to add the sequence-independent high-resolution details. This approach offers several advantages and can overcome key limitations of the SoTA in this domain. Compared to existing image sequence generation models, our method achieves superior synthesis quality and improved coherence across sequences. It also delivers high-fidelity generation of arbitrary-length sequences and increased efficiency in inference time and training data usage. Furthermore, our straightforward formulation enables our method to generalize effectively across diverse data domains, which typically require additional priors and supervision to model in a generative context. Our method consistently outperforms SoTA in quality and inference speed (at least twice-as-fast) across datasets. --- ### GriDiT: Factorized Grid-Based Diffusion for Efficient Long Image Sequence Generation Authors: Snehal Singh Tomar, Alexandros Graikos, Arjun Krishna, Dimitris Samaras, Klaus Mueller Published: 2025-12-24 Categories: cs.CV arXiv: https://arxiv.org/abs/2512.21276v1 Modern deep learning methods typically treat image sequences as large tensors of sequentially stacked frames. However, is this straightforward representation ideal given the current state-of-the-art (SoTA)? In this work, we address this question in the context of generative models and aim to devise a more effective way of modeling image sequence data. Observing the inefficiencies and bottlenecks of current SoTA image sequence generation methods, we showcase that rather than working with large tensors, we can improve the generation process by factorizing it into first generating the coarse sequence at low resolution and then refining the individual frames at high resolution. We train a generative model solely on grid images comprising subsampled frames. Yet, we learn to generate image sequences, using the strong self-attention mechanism of the Diffusion Transformer (DiT) to capture correlations between frames. In effect, our formulation extends a 2D image generator to operate as a low-resolution 3D image-sequence generator without introducing any architectural modifications. Subsequently, we super-resolve each frame individually to add the sequence-independent high-resolution details. This approach offers several advantages and can overcome key limitations of the SoTA in this domain. Compared to existing image sequence generation models, our method achieves superior synthesis quality and improved coherence across sequences. It also delivers high-fidelity generation of arbitrary-length sequences and increased efficiency in inference time and training data usage. Furthermore, our straightforward formulation enables our method to generalize effectively across diverse data domains, which typically require additional priors and supervision to model in a generative context. Our method consistently outperforms SoTA in quality and inference speed (at least twice-as-fast) across datasets. --- ### ACD: Direct Conditional Control for Video Diffusion Models via Attention Supervision Authors: Weiqi Li, Zehao Zhang, Liang Lin, Guangrun Wang Published: 2025-12-24 Categories: cs.CV arXiv: https://arxiv.org/abs/2512.21268v1 Controllability is a fundamental requirement in video synthesis, where accurate alignment with conditioning signals is essential. Existing classifier-free guidance methods typically achieve conditioning indirectly by modeling the joint distribution of data and conditions, which often results in limited controllability over the specified conditions. Classifier-based guidance enforces conditions through an external classifier, but the model may exploit this mechanism to raise the classifier score without genuinely satisfying the intended condition, resulting in adversarial artifacts and limited effective controllability. In this paper, we propose Attention-Conditional Diffusion (ACD), a novel framework for direct conditional control in video diffusion models via attention supervision. By aligning the model's attention maps with external control signals, ACD achieves better controllability. To support this, we introduce a sparse 3D-aware object layout as an efficient conditioning signal, along with a dedicated Layout ControlNet and an automated annotation pipeline for scalable layout integration. Extensive experiments on benchmark video generation datasets demonstrate that ACD delivers superior alignment with conditioning inputs while preserving temporal coherence and visual fidelity, establishing an effective paradigm for conditional video synthesis. --- ### AnyAD: Unified Any-Modality Anomaly Detection in Incomplete Multi-Sequence MRI Authors: Changwei Wu, Yifei Chen, Yuxin Du, Mingxuan Liu, Jinying Zong, Beining Wu, Jie Dong, Feiwei Qin, Yunkang Cao, Qiyuan Tian Published: 2025-12-24 Categories: cs.CV arXiv: https://arxiv.org/abs/2512.21264v1 Reliable anomaly detection in brain MRI remains challenging due to the scarcity of annotated abnormal cases and the frequent absence of key imaging modalities in real clinical workflows. Existing single-class or multi-class anomaly detection (AD) models typically rely on fixed modality configurations, require repetitive training, or fail to generalize to unseen modality combinations, limiting their clinical scalability. In this work, we present a unified Any-Modality AD framework that performs robust anomaly detection and localization under arbitrary MRI modality availability. The framework integrates a dual-pathway DINOv2 encoder with a feature distribution alignment mechanism that statistically aligns incomplete-modality features with full-modality representations, enabling stable inference even with severe modality dropout. To further enhance semantic consistency, we introduce an Intrinsic Normal Prototypes (INPs) extractor and an INP-guided decoder that reconstruct only normal anatomical patterns while naturally amplifying abnormal deviations. Through randomized modality masking and indirect feature completion during training, the model learns to adapt to all modality configurations without re-training. Extensive experiments on BraTS2018, MU-Glioma-Post, and Pretreat-MetsToBrain-Masks demonstrate that our approach consistently surpasses state-of-the-art industrial and medical AD baselines across 7 modality combinations, achieving superior generalization. This study establishes a scalable paradigm for multimodal medical AD under real-world, imperfect modality conditions. Our source code is available at https://github.com/wuchangw/AnyAD. --- ### ReaSeq: Unleashing World Knowledge via Reasoning for Sequential Modeling Authors: Jiakai Tang, Chuan Wang, Gaoming Yang, Han Wu, Jiahao Yu, Jian Wu, Jianwu Hu, Junjun Zheng, Longbin Li, Shuwen Xiao, Xiangheng Kong, Yeqiu Yang et al. Published: 2025-12-24 Categories: cs.IR, cs.CL arXiv: https://arxiv.org/abs/2512.21257v2 Industrial recommender systems face two fundamental limitations under the log-driven paradigm: (1) knowledge poverty in ID-based item representations that causes brittle interest modeling under data sparsity, and (2) systemic blindness to beyond-log user interests that constrains model performance within platform boundaries. These limitations stem from an over-reliance on shallow interaction statistics and close-looped feedback while neglecting the rich world knowledge about product semantics and cross-domain behavioral patterns that Large Language Models have learned from vast corpora. To address these challenges, we introduce ReaSeq, a reasoning-enhanced framework that leverages world knowledge in Large Language Models to address both limitations through explicit and implicit reasoning. Specifically, ReaSeq employs explicit Chain-of-Thought reasoning via multi-agent collaboration to distill structured product knowledge into semantically enriched item representations, and latent reasoning via Diffusion Large Language Models to infer plausible beyond-log behaviors. Deployed on Taobao's ranking system serving hundreds of millions of users, ReaSeq achieves substantial gains: >6.0% in IPV and CTR, >2.9% in Orders, and >2.5% in GMV, validating the effectiveness of world-knowledge-enhanced reasoning over purely log-driven approaches. --- ### ReaSeq: Unleashing World Knowledge via Reasoning for Sequential Modeling Authors: Chuan Wang, Gaoming Yang, Han Wu, Jiakai Tang, Jiahao Yu, Jian Wu, Jianwu Hu, Junjun Zheng, Shuwen Xiao, Yeqiu Yang, Yuning Jiang, Ahjol Nurlanbek et al. Published: 2025-12-24 Categories: cs.IR, cs.CL arXiv: https://arxiv.org/abs/2512.21257v1 Industrial recommender systems face two fundamental limitations under the log-driven paradigm: (1) knowledge poverty in ID-based item representations that causes brittle interest modeling under data sparsity, and (2) systemic blindness to beyond-log user interests that constrains model performance within platform boundaries. These limitations stem from an over-reliance on shallow interaction statistics and close-looped feedback while neglecting the rich world knowledge about product semantics and cross-domain behavioral patterns that Large Language Models have learned from vast corpora. To address these challenges, we introduce ReaSeq, a reasoning-enhanced framework that leverages world knowledge in Large Language Models to address both limitations through explicit and implicit reasoning. Specifically, ReaSeq employs explicit Chain-of-Thought reasoning via multi-agent collaboration to distill structured product knowledge into semantically enriched item representations, and latent reasoning via Diffusion Large Language Models to infer plausible beyond-log behaviors. Deployed on Taobao's ranking system serving hundreds of millions of users, ReaSeq achieves substantial gains: >6.0% in IPV and CTR, >2.9% in Orders, and >2.5% in GMV, validating the effectiveness of world-knowledge-enhanced reasoning over purely log-driven approaches. --- ### DreaMontage: Arbitrary Frame-Guided One-Shot Video Generation Authors: Jiawei Liu, Junqiao Li, Jiangfan Deng, Gen Li, Siyu Zhou, Zetao Fang, Shanshan Lao, Zengde Deng, Jianing Zhu, Tingting Ma, Jiayi Li, Yunqiu Wang et al. Published: 2025-12-24 Categories: cs.CV arXiv: https://arxiv.org/abs/2512.21252v2 The "one-shot" technique represents a distinct and sophisticated aesthetic in filmmaking. However, its practical realization is often hindered by prohibitive costs and complex real-world constraints. Although emerging video generation models offer a virtual alternative, existing approaches typically rely on naive clip concatenation, which frequently fails to maintain visual smoothness and temporal coherence. In this paper, we introduce DreaMontage, a comprehensive framework designed for arbitrary frame-guided generation, capable of synthesizing seamless, expressive, and long-duration one-shot videos from diverse user-provided inputs. To achieve this, we address the challenge through three primary dimensions. (i) We integrate a lightweight intermediate-conditioning mechanism into the DiT architecture. By employing an Adaptive Tuning strategy that effectively leverages base training data, we unlock robust arbitrary-frame control capabilities. (ii) To enhance visual fidelity and cinematic expressiveness, we curate a high-quality dataset and implement a Visual Expression SFT stage. In addressing critical issues such as subject motion rationality and transition smoothness, we apply a Tailored DPO scheme, which significantly improves the success rate and usability of the generated content. (iii) To facilitate the production of extended sequences, we design a Segment-wise Auto-Regressive (SAR) inference strategy that operates in a memory-efficient manner. Extensive experiments demonstrate that our approach achieves visually striking and seamlessly coherent one-shot effects while maintaining computational efficiency, empowering users to transform fragmented visual materials into vivid, cohesive one-shot cinematic experiences. --- ### DreaMontage: Arbitrary Frame-Guided One-Shot Video Generation Authors: Jiawei Liu, Junqiao Li, Jiangfan Deng, Gen Li, Siyu Zhou, Zetao Fang, Shanshan Lao, Zengde Deng, Jianing Zhu, Tingting Ma, Jiayi Li, Yunqiu Wang et al. Published: 2025-12-24 Categories: cs.CV arXiv: https://arxiv.org/abs/2512.21252v1 The "one-shot" technique represents a distinct and sophisticated aesthetic in filmmaking. However, its practical realization is often hindered by prohibitive costs and complex real-world constraints. Although emerging video generation models offer a virtual alternative, existing approaches typically rely on naive clip concatenation, which frequently fails to maintain visual smoothness and temporal coherence. In this paper, we introduce DreaMontage, a comprehensive framework designed for arbitrary frame-guided generation, capable of synthesizing seamless, expressive, and long-duration one-shot videos from diverse user-provided inputs. To achieve this, we address the challenge through three primary dimensions. (i) We integrate a lightweight intermediate-conditioning mechanism into the DiT architecture. By employing an Adaptive Tuning strategy that effectively leverages base training data, we unlock robust arbitrary-frame control capabilities. (ii) To enhance visual fidelity and cinematic expressiveness, we curate a high-quality dataset and implement a Visual Expression SFT stage. In addressing critical issues such as subject motion rationality and transition smoothness, we apply a Tailored DPO scheme, which significantly improves the success rate and usability of the generated content. (iii) To facilitate the production of extended sequences, we design a Segment-wise Auto-Regressive (SAR) inference strategy that operates in a memory-efficient manner. Extensive experiments demonstrate that our approach achieves visually striking and seamlessly coherent one-shot effects while maintaining computational efficiency, empowering users to transform fragmented visual materials into vivid, cohesive one-shot cinematic experiences. --- ### Learning Factors in AI-Augmented Education: A Comparative Study of Middle and High School Students Authors: Gaia Ebli, Bianca Raimondi, Maurizio Gabbrielli Published: 2025-12-24 Categories: cs.HC, cs.AI arXiv: https://arxiv.org/abs/2512.21246v1 The increasing integration of AI tools in education has led prior research to explore their impact on learning processes. Nevertheless, most existing studies focus on higher education and conventional instructional contexts, leaving open questions about how key learning factors are related in AI-mediated learning environments and how these relationships may vary across different age groups. Addressing these gaps, our work investigates whether four critical learning factors, experience, clarity, comfort, and motivation, maintain coherent interrelationships in AI-augmented educational settings, and how the structure of these relationships differs between middle and high school students. The study was conducted in authentic classroom contexts where students interacted with AI tools as part of programming learning activities to collect data on the four learning factors and students' perceptions. Using a multimethod quantitative analysis, which combined correlation analysis and text mining, we revealed markedly different dimensional structures between the two age groups. Middle school students exhibit strong positive correlations across all dimensions, indicating holistic evaluation patterns whereby positive perceptions in one dimension generalise to others. In contrast, high school students show weak or near-zero correlations between key dimensions, suggesting a more differentiated evaluation process in which dimensions are assessed independently. These findings reveal that perception dimensions actively mediate AI-augmented learning and that the developmental stage moderates their interdependencies. This work establishes a foundation for the development of AI integration strategies that respond to learners' developmental levels and account for age-specific dimensional structures in student-AI interactions. --- ### LookPlanGraph: Embodied Instruction Following Method with VLM Graph Augmentation Authors: Anatoly O. Onishchenko, Alexey K. Kovalev, Aleksandr I. Panov Published: 2025-12-24 Categories: cs.RO, cs.AI, cs.LG arXiv: https://arxiv.org/abs/2512.21243v1 Methods that use Large Language Models (LLM) as planners for embodied instruction following tasks have become widespread. To successfully complete tasks, the LLM must be grounded in the environment in which the robot operates. One solution is to use a scene graph that contains all the necessary information. Modern methods rely on prebuilt scene graphs and assume that all task-relevant information is available at the start of planning. However, these approaches do not account for changes in the environment that may occur between the graph construction and the task execution. We propose LookPlanGraph - a method that leverages a scene graph composed of static assets and object priors. During plan execution, LookPlanGraph continuously updates the graph with relevant objects, either by verifying existing priors or discovering new entities. This is achieved by processing the agents egocentric camera view using a Vision Language Model. We conducted experiments with changed object positions VirtualHome and OmniGibson simulated environments, demonstrating that LookPlanGraph outperforms methods based on predefined static scene graphs. To demonstrate the practical applicability of our approach, we also conducted experiments in a real-world setting. Additionally, we introduce the GraSIF (Graph Scenes for Instruction Following) dataset with automated validation framework, comprising 514 tasks drawn from SayPlan Office, BEHAVIOR-1K, and VirtualHome RobotHow. Project page available at https://lookplangraph.github.io . --- ### Improving the Convergence Rate of Ray Search Optimization for Query-Efficient Hard-Label Attacks Authors: Xinjie Xu, Shuyu Cheng, Dongwei Xu, Qi Xuan, Chen Ma Published: 2025-12-24 Categories: cs.LG, cs.AI, cs.CR, cs.CV arXiv: https://arxiv.org/abs/2512.21241v1 In hard-label black-box adversarial attacks, where only the top-1 predicted label is accessible, the prohibitive query complexity poses a major obstacle to practical deployment. In this paper, we focus on optimizing a representative class of attacks that search for the optimal ray direction yielding the minimum $\ell_2$-norm perturbation required to move a benign image into the adversarial region. Inspired by Nesterov's Accelerated Gradient (NAG), we propose a momentum-based algorithm, ARS-OPT, which proactively estimates the gradient with respect to a future ray direction inferred from accumulated momentum. We provide a theoretical analysis of its convergence behavior, showing that ARS-OPT enables more accurate directional updates and achieves faster, more stable optimization. To further accelerate convergence, we incorporate surrogate-model priors into ARS-OPT's gradient estimation, resulting in PARS-OPT with enhanced performance. The superiority of our approach is supported by theoretical guarantees under standard assumptions. Extensive experiments on ImageNet and CIFAR-10 demonstrate that our method surpasses 13 state-of-the-art approaches in query efficiency. --- ### Assessing the Software Security Comprehension of Large Language Models Authors: Mohammed Latif Siddiq, Natalie Sekerak, Antonio Karam, Maria Leal, Arvin Islam-Gomes, Joanna C. S. Santos Published: 2025-12-24 Categories: cs.SE, cs.CR, cs.LG arXiv: https://arxiv.org/abs/2512.21238v1 Large language models (LLMs) are increasingly used in software development, but their level of software security expertise remains unclear. This work systematically evaluates the security comprehension of five leading LLMs: GPT-4o-Mini, GPT-5-Mini, Gemini-2.5-Flash, Llama-3.1, and Qwen-2.5, using Blooms Taxonomy as a framework. We assess six cognitive dimensions: remembering, understanding, applying, analyzing, evaluating, and creating. Our methodology integrates diverse datasets, including curated multiple-choice questions, vulnerable code snippets (SALLM), course assessments from an Introduction to Software Security course, real-world case studies (XBOW), and project-based creation tasks from a Secure Software Engineering course. Results show that while LLMs perform well on lower-level cognitive tasks such as recalling facts and identifying known vulnerabilities, their performance degrades significantly on higher-order tasks that require reasoning, architectural evaluation, and secure system creation. Beyond reporting aggregate accuracy, we introduce a software security knowledge boundary that identifies the highest cognitive level at which a model consistently maintains reliable performance. In addition, we identify 51 recurring misconception patterns exhibited by LLMs across Blooms levels. --- ### SegMo: Segment-aligned Text to 3D Human Motion Generation Authors: Bowen Dang, Lin Wu, Xiaohang Yang, Zheng Yuan, Zhixiang Chen Published: 2025-12-24 Categories: cs.CV arXiv: https://arxiv.org/abs/2512.21237v1 Generating 3D human motions from textual descriptions is an important research problem with broad applications in video games, virtual reality, and augmented reality. Recent methods align the textual description with human motion at the sequence level, neglecting the internal semantic structure of modalities. However, both motion descriptions and motion sequences can be naturally decomposed into smaller and semantically coherent segments, which can serve as atomic alignment units to achieve finer-grained correspondence. Motivated by this, we propose SegMo, a novel Segment-aligned text-conditioned human Motion generation framework to achieve fine-grained text-motion alignment. Our framework consists of three modules: (1) Text Segment Extraction, which decomposes complex textual descriptions into temporally ordered phrases, each representing a simple atomic action; (2) Motion Segment Extraction, which partitions complete motion sequences into corresponding motion segments; and (3) Fine-grained Text-Motion Alignment, which aligns text and motion segments with contrastive learning. Extensive experiments demonstrate that SegMo improves the strong baseline on two widely used datasets, achieving an improved TOP 1 score of 0.553 on the HumanML3D test set. Moreover, thanks to the learned shared embedding space for text and motion segments, SegMo can also be applied to retrieval-style tasks such as motion grounding and motion-to-text retrieval. --- ### Casting a SPELL: Sentence Pairing Exploration for LLM Limitation-breaking Authors: Yifan Huang, Xiaojun Jia, Wenbo Guo, Yuqiang Sun, Yihao Huang, Chong Wang, Yang Liu Published: 2025-12-24 Categories: cs.CR, cs.AI, cs.SE arXiv: https://arxiv.org/abs/2512.21236v1 Large language models (LLMs) have revolutionized software development through AI-assisted coding tools, enabling developers with limited programming expertise to create sophisticated applications. However, this accessibility extends to malicious actors who may exploit these powerful tools to generate harmful software. Existing jailbreaking research primarily focuses on general attack scenarios against LLMs, with limited exploration of malicious code generation as a jailbreak target. To address this gap, we propose SPELL, a comprehensive testing framework specifically designed to evaluate the weakness of security alignment in malicious code generation. Our framework employs a time-division selection strategy that systematically constructs jailbreaking prompts by intelligently combining sentences from a prior knowledge dataset, balancing exploration of novel attack patterns with exploitation of successful techniques. Extensive evaluation across three advanced code models (GPT-4.1, Claude-3.5, and Qwen2.5-Coder) demonstrates SPELL's effectiveness, achieving attack success rates of 83.75%, 19.38%, and 68.12% respectively across eight malicious code categories. The generated prompts successfully produce malicious code in real-world AI development tools such as Cursor, with outputs confirmed as malicious by state-of-the-art detection systems at rates exceeding 73%. These findings reveal significant security gaps in current LLM implementations and provide valuable insights for improving AI safety alignment in code generation applications. --- ### MiST: Understanding the Role of Mid-Stage Scientific Training in Developing Chemical Reasoning Models Authors: Andres M Bran, Tong Xie, Shai Pranesh, Jeffrey Meng, Xuan Vu Nguyen, Jeremy Goumaz, David Ming Segura, Ruizhi Xu, Dongzhan Zhou, Wenjie Zhang, Bram Hoex, Philippe Schwaller Published: 2025-12-24 Categories: cs.LG, cond-mat.mtrl-sci arXiv: https://arxiv.org/abs/2512.21231v3 Large Language Models can develop reasoning capabilities through online fine-tuning with rule-based rewards. However, recent studies reveal a critical constraint: reinforcement learning succeeds only when the base model already assigns non-negligible probability to correct answers -- a property we term 'latent solvability'. This work investigates the emergence of chemical reasoning capabilities and what these prerequisites mean for chemistry. We identify two necessary conditions for RL-based chemical reasoning: 1) Symbolic competence, and 2) Latent chemical knowledge. We propose mid-stage scientific training (MiST): a set of mid-stage training techniques to satisfy these, including data-mixing with SMILES/CIF-aware pre-processing, continued pre-training on 2.9B tokens, and supervised fine-tuning on 1B tokens. These steps raise the latent-solvability score on 3B and 7B models by up to 1.8x, and enable RL to lift top-1 accuracy from 10.9 to 63.9% on organic reaction naming, and from 40.6 to 67.4% on inorganic material generation. Similar results are observed for other challenging chemical tasks, while producing interpretable reasoning traces. Our results define clear prerequisites for chemical reasoning training and highlight the broader role of mid-stage training in unlocking reasoning capabilities. --- ### MiST: Understanding the Role of Mid-Stage Scientific Training in Developing Chemical Reasoning Models Authors: Andres M Bran, Tong Xie, Shai Pranesh, Jeffrey Meng, Xuan Vu Nguyen, Jeremy Goumaz, David Ming Segura, Ruizhi Xu, Dongzhan Zhou, Wenjie Zhang, Bram Hoex, Philippe Schwaller Published: 2025-12-24 Categories: cs.LG, cond-mat.mtrl-sci arXiv: https://arxiv.org/abs/2512.21231v1 Large Language Models can develop reasoning capabilities through online fine-tuning with rule-based rewards. However, recent studies reveal a critical constraint: reinforcement learning succeeds only when the base model already assigns non-negligible probability to correct answers -- a property we term 'latent solvability'. This work investigates the emergence of chemical reasoning capabilities and what these prerequisites mean for chemistry. We identify two necessary conditions for RL-based chemical reasoning: 1) Symbolic competence, and 2) Latent chemical knowledge. We propose mid-stage scientific training (MiST): a set of mid-stage training techniques to satisfy these, including data-mixing with SMILES/CIF-aware pre-processing, continued pre-training on 2.9B tokens, and supervised fine-tuning on 1B tokens. These steps raise the latent-solvability score on 3B and 7B models by up to 1.8x, and enable RL to lift top-1 accuracy from 10.9 to 63.9% on organic reaction naming, and from 40.6 to 67.4% on inorganic material generation. Similar results are observed for other challenging chemical tasks, while producing interpretable reasoning traces. Our results define clear prerequisites for chemical reasoning training and highlight the broader role of mid-stage training in unlocking reasoning capabilities. --- ### PhononBench:A Large-Scale Phonon-Based Benchmark for Dynamical Stability in Crystal Generation Authors: Xiao-Qi Han, Ze-Feng Gao, Wen-Kao Li, Peng-Jie Guo, Zhong-Yi Lu Published: 2025-12-24 Categories: cond-mat.mtrl-sci, cs.AI arXiv: https://arxiv.org/abs/2512.21227v3 In recent years, generative artificial intelligence has made significant advances in the design of crystalline materials, giving rise to approaches based on graph neural networks, diffusion models, and large language models. Existing evaluations commonly follow the stability-uniqueness-novelty (S.U.N.) framework, where stability is primarily assessed using thermodynamic criteria, which do not fully capture the dynamical stability essential for a material's practical existence. Dynamical stability is a key determinant of whether a material can be synthesized and persist, with phonon spectrum calculations serving as the standard for its evaluation. However, the high computational cost of such calculations has prevented large-scale assessment of dynamical stability in generated crystals. In this work, we introduce PhononBench, the first large-scale benchmark for dynamical stability in AI-generated crystals. Leveraging the recently developed MatterSim interatomic potential, which achieves density-functional-theory (DFT)-level accuracy in phonon predictions across more than 10,000 materials, PhononBench enables efficient phonon calculations and dynamical-stability analysis for 133,838 crystal structures generated by 7 leading crystal generation models. PhononBench reveals a widespread limitation of current generative models: unless otherwise specified, all reported dynamical-stability metrics are evaluated at a phonon-frequency threshold of -0.1 THz, with the average dynamical-stability rate across all generated structures being only 32.15%, and the top-performing model, MatterGen, reaching just 45.05%.In addition, we identify 32,995 crystal structures that are phonon-stable across the entire Brillouin zone under a strict threshold of -0.001 THz. In addition, a web-based service is accessible at http://phononbench.cn/, enabling minute-level ultra-fast phonon predictions. --- ### PhononBench:A Large-Scale Phonon-Based Benchmark for Dynamical Stability in Crystal Generation Authors: Xiao-Qi Han, Ze-Feng Gao, Peng-Jie Guo, Zhong-Yi Lu Published: 2025-12-24 Categories: cond-mat.mtrl-sci, cs.AI arXiv: https://arxiv.org/abs/2512.21227v1 In this work, we introduce PhononBench, the first large-scale benchmark for dynamical stability in AI-generated crystals. Leveraging the recently developed MatterSim interatomic potential, which achieves DFT-level accuracy in phonon predictions across more than 10,000 materials, PhononBench enables efficient large-scale phonon calculations and dynamical-stability analysis for 108,843 crystal structures generated by six leading crystal generation models. PhononBench reveals a widespread limitation of current generative models in ensuring dynamical stability: the average dynamical-stability rate across all generated structures is only 25.83%, with the top-performing model, MatterGen, reaching just 41.0%. Further case studies show that in property-targeted generation-illustrated here by band-gap conditioning with MatterGen--the dynamical-stability rate remains as low as 23.5% even at the optimal band-gap condition of 0.5 eV. In space-group-controlled generation, higher-symmetry crystals exhibit better stability (e.g., cubic systems achieve rates up to 49.2%), yet the average stability across all controlled generations is still only 34.4%. An important additional outcome of this study is the identification of 28,119 crystal structures that are phonon-stable across the entire Brillouin zone, providing a substantial pool of reliable candidates for future materials exploration. By establishing the first large-scale dynamical-stability benchmark, this work systematically highlights the current limitations of crystal generation models and offers essential evaluation criteria and guidance for their future development toward the design and discovery of physically viable materials. All model-generated crystal structures, phonon calculation results, and the high-throughput evaluation workflows developed in PhononBench will be openly released at https://github.com/xqh19970407/PhononBench --- ### Leveraging Lightweight Entity Extraction for Scalable Event-Based Image Retrieval Authors: Dao Sy Duy Minh, Huynh Trung Kiet, Nguyen Lam Phu Quy, Phu-Hoa Pham, Tran Chi Nguyen Published: 2025-12-24 Categories: cs.CV, cs.AI arXiv: https://arxiv.org/abs/2512.21221v1 Retrieving images from natural language descriptions is a core task at the intersection of computer vision and natural language processing, with wide-ranging applications in search engines, media archiving, and digital content management. However, real-world image-text retrieval remains challenging due to vague or context-dependent queries, linguistic variability, and the need for scalable solutions. In this work, we propose a lightweight two-stage retrieval pipeline that leverages event-centric entity extraction to incorporate temporal and contextual signals from real-world captions. The first stage performs efficient candidate filtering using BM25 based on salient entities, while the second stage applies BEiT-3 models to capture deep multimodal semantics and rerank the results. Evaluated on the OpenEvents v1 benchmark, our method achieves a mean average precision of 0.559, substantially outperforming prior baselines. These results highlight the effectiveness of combining event-guided filtering with long-text vision-language modeling for accurate and efficient retrieval in complex, real-world scenarios. Our code is available at https://github.com/PhamPhuHoa-23/Event-Based-Image-Retrieval --- ### RoboSafe: Safeguarding Embodied Agents via Executable Safety Logic Authors: Le Wang, Zonghao Ying, Xiao Yang, Quanchen Zou, Zhenfei Yin, Tianlin Li, Jian Yang, Yaodong Yang, Aishan Liu, Xianglong Liu Published: 2025-12-24 Categories: cs.AI, cs.CV, cs.RO arXiv: https://arxiv.org/abs/2512.21220v2 Embodied agents powered by vision-language models (VLMs) are increasingly capable of executing complex real-world tasks, yet they remain vulnerable to hazardous instructions that may trigger unsafe behaviors. Runtime safety guardrails, which intercept hazardous actions during task execution, offer a promising solution due to their flexibility. However, existing defenses often rely on static rule filters or prompt-level control, which struggle to address implicit risks arising in dynamic, temporally dependent, and context-rich environments. To address this, we propose RoboSafe, a hybrid reasoning runtime safeguard for embodied agents through executable predicate-based safety logic. RoboSafe integrates two complementary reasoning processes on a Hybrid Long-Short Safety Memory. We first propose a Backward Reflective Reasoning module that continuously revisits recent trajectories in short-term memory to infer temporal safety predicates and proactively triggers replanning when violations are detected. We then propose a Forward Predictive Reasoning module that anticipates upcoming risks by generating context-aware safety predicates from the long-term safety memory and the agent's multimodal observations. Together, these components form an adaptive, verifiable safety logic that is both interpretable and executable as code. Extensive experiments across multiple agents demonstrate that RoboSafe substantially reduces hazardous actions (-36.8% risk occurrence) compared with leading baselines, while maintaining near-original task performance. Real-world evaluations on physical robotic arms further confirm its practicality. Code will be released upon acceptance. --- ### RoboSafe: Safeguarding Embodied Agents via Executable Safety Logic Authors: Le Wang, Zonghao Ying, Xiao Yang, Quanchen Zou, Zhenfei Yin, Tianlin Li, Jian Yang, Yaodong Yang, Aishan Liu, Xianglong Liu Published: 2025-12-24 Categories: cs.AI, cs.CV, cs.RO arXiv: https://arxiv.org/abs/2512.21220v1 Embodied agents powered by vision-language models (VLMs) are increasingly capable of executing complex real-world tasks, yet they remain vulnerable to hazardous instructions that may trigger unsafe behaviors. Runtime safety guardrails, which intercept hazardous actions during task execution, offer a promising solution due to their flexibility. However, existing defenses often rely on static rule filters or prompt-level control, which struggle to address implicit risks arising in dynamic, temporally dependent, and context-rich environments. To address this, we propose RoboSafe, a hybrid reasoning runtime safeguard for embodied agents through executable predicate-based safety logic. RoboSafe integrates two complementary reasoning processes on a Hybrid Long-Short Safety Memory. We first propose a Backward Reflective Reasoning module that continuously revisits recent trajectories in short-term memory to infer temporal safety predicates and proactively triggers replanning when violations are detected. We then propose a Forward Predictive Reasoning module that anticipates upcoming risks by generating context-aware safety predicates from the long-term safety memory and the agent's multimodal observations. Together, these components form an adaptive, verifiable safety logic that is both interpretable and executable as code. Extensive experiments across multiple agents demonstrate that RoboSafe substantially reduces hazardous actions (-36.8% risk occurrence) compared with leading baselines, while maintaining near-original task performance. Real-world evaluations on physical robotic arms further confirm its practicality. Code will be released upon acceptance. --- ### Latent Implicit Visual Reasoning Authors: Kelvin Li, Chuyi Shang, Leonid Karlinsky, Rogerio Feris, Trevor Darrell, Roei Herzig Published: 2025-12-24 Categories: cs.CV arXiv: https://arxiv.org/abs/2512.21218v2 While Large Multimodal Models (LMMs) have made significant progress, they remain largely text-centric, relying on language as their core reasoning modality. As a result, they are limited in their ability to handle reasoning tasks that are predominantly visual. Recent approaches have sought to address this by supervising intermediate visual steps with helper images, depth maps, or image crops. However, these strategies impose restrictive priors on what "useful" visual abstractions look like, add heavy annotation costs, and struggle to generalize across tasks. To address this critical limitation, we propose Latent Implicit Visual Reasoning (LIVR), a task-agnostic mechanism that trains LMMs to discover and use latent visual reasoning tokens without explicit intermediate supervision. These tokens attend globally and re-encode the image in a task-adaptive way, enabling the model to extract relevant visual information without hand-crafted supervision. LIVR consistently outperforms direct supervised fine-tuning across diverse vision-centric tasks and multiple LMM backbones. In broader comparisons, LIVR remains competitive with or outperforms prior text-based and explicit-visual-intermediate reasoning methods, while requiring no additional intermediate supervision such as helper images, bounding boxes, image crops, depth maps, or chain-of-thought annotations. Our project page can be found here: https://www.chuyishang.com/livr/ --- ### Latent Implicit Visual Reasoning Authors: Kelvin Li, Chuyi Shang, Leonid Karlinsky, Rogerio Feris, Trevor Darrell, Roei Herzig Published: 2025-12-24 Categories: cs.CV arXiv: https://arxiv.org/abs/2512.21218v1 While Large Multimodal Models (LMMs) have made significant progress, they remain largely text-centric, relying on language as their core reasoning modality. As a result, they are limited in their ability to handle reasoning tasks that are predominantly visual. Recent approaches have sought to address this by supervising intermediate visual steps with helper images, depth maps, or image crops. However, these strategies impose restrictive priors on what "useful" visual abstractions look like, add heavy annotation costs, and struggle to generalize across tasks. To address this critical limitation, we propose a task-agnostic mechanism that trains LMMs to discover and use visual reasoning tokens without explicit supervision. These tokens attend globally and re-encode the image in a task-adaptive way, enabling the model to extract relevant visual information without hand-crafted supervision. Our approach outperforms direct fine-tuning and achieves state-of-the-art results on a diverse range of vision-centric tasks -- including those where intermediate abstractions are hard to specify -- while also generalizing to multi-task instruction tuning. --- ### Causal-driven attribution (CDA): Estimating channel influence without user-level data Authors: Georgios Filippou, Boi Mai Quach, Diana Lenghel, Arthur White, Ashish Kumar Jha Published: 2025-12-24 Categories: stat.ML, cs.LG arXiv: https://arxiv.org/abs/2512.21211v1 Attribution modelling lies at the heart of marketing effectiveness, yet most existing approaches depend on user-level path data, which are increasingly inaccessible due to privacy regulations and platform restrictions. This paper introduces a Causal-Driven Attribution (CDA) framework that infers channel influence using only aggregated impression-level data, avoiding any reliance on user identifiers or click-path tracking. CDA integrates temporal causal discovery (using PCMCI) with causal effect estimation via a Structural Causal Model to recover directional channel relationships and quantify their contributions to conversions. Using large-scale synthetic data designed to replicate real marketing dynamics, we show that CDA achieves an average relative RMSE of 9.50% when given the true causal graph, and 24.23% when using the predicted graph, demonstrating strong accuracy under correct structure and meaningful signal recovery even under structural uncertainty. CDA captures cross-channel interdependencies while providing interpretable, privacy-preserving attribution insights, offering a scalable and future-proof alternative to traditional path-based models. --- ### Human Motion Estimation with Everyday Wearables Authors: Siqi Zhu, Yixuan Li, Junfu Li, Qi Wu, Zan Wang, Haozhe Ma, Wei Liang Published: 2025-12-24 Categories: cs.CV arXiv: https://arxiv.org/abs/2512.21209v1 While on-body device-based human motion estimation is crucial for applications such as XR interaction, existing methods often suffer from poor wearability, expensive hardware, and cumbersome calibration, which hinder their adoption in daily life. To address these challenges, we present EveryWear, a lightweight and practical human motion capture approach based entirely on everyday wearables: a smartphone, smartwatch, earbuds, and smart glasses equipped with one forward-facing and two downward-facing cameras, requiring no explicit calibration before use. We introduce Ego-Elec, a 9-hour real-world dataset covering 56 daily activities across 17 diverse indoor and outdoor environments, with ground-truth 3D annotations provided by the motion capture (MoCap), to facilitate robust research and benchmarking in this direction. Our approach employs a multimodal teacher-student framework that integrates visual cues from egocentric cameras with inertial signals from consumer devices. By training directly on real-world data rather than synthetic data, our model effectively eliminates the sim-to-real gap that constrains prior work. Experiments demonstrate that our method outperforms baseline models, validating its effectiveness for practical full-body motion estimation. --- ### A Learning Stability Profile for Finite-Dimensional Learning Dynamics Authors: Ronald Katende Published: 2025-12-24 Categories: cs.LG, math.DS, math.OC arXiv: https://arxiv.org/abs/2512.21208v3 We develop a finite-dimensional sensitivity framework for studying stability in learning systems whose states include representations, parameters, and update variables. The central object is the \emph{Learning Stability Profile}, a collection of directional sensitivity operators that records how perturbations in inputs, parameter initialization, and update mechanisms propagate along a specified learning trajectory. The main result is a Lyapunov criterion for controlling this profile. Under explicit regularity, coercivity, and dissipation assumptions, an incremental Lyapunov energy yields uniform or exponentially decaying bounds on the associated linearized transition operators. The result is stated as a sufficient stability criterion, not as an unconditional converse theorem. The framework also distinguishes terminal decay, profile-wise boundedness, and subexponential growth, avoiding the identification of nonpositive growth exponents with uniform boundedness. The profile is then specialized to several standard learning mechanisms. Spectral bounds give forward sensitivity estimates for feedforward networks. Dissipativity and step-size restrictions give stability bounds for residual architectures. Mean-square contraction assumptions yield parameter and update sensitivity bounds for stochastic gradient methods. Locally Lipschitz systems, including piecewise-linear networks, proximal maps, projected updates, and recurrent or state-space recursions, are handled through Clarke generalized Jacobians and variational Lyapunov inequalities. The resulting framework provides a common stability language for architecture, optimization, stochasticity, and nonsmoothness. Its role is structural: it organizes known stability mechanisms within one perturbation calculus while keeping the hypotheses needed for each guarantee explicit. --- ### Analytic and Variational Stability of Deep Learning Systems Authors: Ronald Katende Published: 2025-12-24 Categories: cs.LG, math.DS, math.OC arXiv: https://arxiv.org/abs/2512.21208v1 We propose a unified analytic and variational framework for studying stability in deep learning systems viewed as coupled representation-parameter dynamics. The central object is the Learning Stability Profile, which tracks the infinitesimal response of representations, parameters, and update mechanisms to perturbations along the learning trajectory. We prove a Fundamental Analytic Stability Theorem showing that uniform boundedness of these stability signatures is equivalent, up to norm equivalence, to the existence of a Lyapunov-type energy that dissipates along the learning flow. In smooth regimes, the framework yields explicit stability exponents linking spectral norms, activation regularity, step sizes, and learning rates to contractivity of the learning dynamics. Classical spectral stability results for feedforward networks, a discrete CFL-type condition for residual architectures, and parametric and temporal stability laws for stochastic gradient methods arise as direct consequences. The theory extends to non-smooth learning systems, including ReLU networks, proximal and projected updates, and stochastic subgradient flows, by replacing classical derivatives with Clarke generalized derivatives and smooth energies with variational Lyapunov functionals. The resulting framework provides a unified dynamical description of stability across architectures and optimization methods, clarifying how architectural and algorithmic choices jointly govern robustness and sensitivity to perturbations. It also provides a foundation for further extensions to continuous-time limits and geometric formulations of learning dynamics. ---