# Research Papers — 2025-10 · bundle page 163 of 217

> Full-text markdown of papers #8101–8150 of 10843 published in 2025-10, newest first. AI-readable mirrors of the full 10843-paper 2025-10 archive (50 papers per file).

Canonical page: https://aipod.dev/papers/archive/2025-10/page/163
[← Previous bundle](https://aipod.dev/papers/archive/2025-10/page/162.md)
[Next bundle →](https://aipod.dev/papers/archive/2025-10/page/164.md)

### Agent Bain vs. Agent McKinsey: A New Text-to-SQL Benchmark for the Business Domain

Authors: Yue Li, Ran Tao, Derek Hommel, Yusuf Denizay Dönder, Sungyong Chang, David Mimno, Unso Eun Seo Jo
Published: 2025-10-08
Categories: cs.CL
arXiv: https://arxiv.org/abs/2510.07309v4

Text-to-SQL benchmarks have traditionally only tested simple data access as a translation task of natural language to SQL queries. But in reality, users tend to ask diverse questions that require more complex responses including data-driven predictions or recommendations. Using the business domain as a motivating example, we introduce CORGI, a new benchmark that expands text-to-SQL to reflect practical database queries encountered by end users. CORGI is composed of synthetic databases inspired by enterprises such as DoorDash, Airbnb, and Lululemon. It provides questions across four increasingly complicated categories of business queries: descriptive, explanatory, predictive, and recommendational. This challenge calls for causal reasoning, temporal forecasting, and strategic recommendation, reflecting multi-level and multi-step agentic intelligence. We find that LLM performance degrades on higher-level questions as question complexity increases. CORGI also introduces and encourages the text-to-SQL community to consider new automatic methods for evaluating open-ended, qualitative responses in data access tasks. Our experiments show that LLMs exhibit an average 33.12% lower success execution rate (SER) on CORGI compared to existing benchmarks such as BIRD, highlighting the substantially higher complexity of real-world business needs. We release the CORGI dataset, an evaluation framework, and a submission website to support future research.

---

### MLE-Smith: Scaling MLE Tasks with Automated Multi-Agent Pipeline

Authors: Rushi Qiang, Yuchen Zhuang, Anikait Singh, Percy Liang, Chao Zhang, Sherry Yang, Bo Dai
Published: 2025-10-08
Categories: cs.LG, cs.AI
arXiv: https://arxiv.org/abs/2510.07307v1

While Language Models (LMs) have made significant progress in automating machine learning engineering (MLE), the acquisition of high-quality MLE training data is significantly constrained. Current MLE benchmarks suffer from low scalability and limited applicability because they rely on static, manually curated tasks, demanding extensive time and manual effort to produce. We introduce MLE-Smith, a fully automated multi-agent pipeline, to transform raw datasets into competition-style MLE challenges through an efficient generate-verify-execute paradigm for scaling MLE tasks with verifiable quality, real-world usability, and rich diversity. The proposed multi-agent pipeline in MLE-Smith drives structured task design and standardized refactoring, coupled with a hybrid verification mechanism that enforces strict structural rules and high-level semantic soundness. It further validates empirical solvability and real-world fidelity through interactive execution. We apply MLE-Smith to 224 of real-world datasets and generate 606 tasks spanning multiple categories, objectives, and modalities, demonstrating that MLE-Smith can work effectively across a wide range of real-world datasets. Evaluation on the generated tasks shows that the performance of eight mainstream and cutting-edge LLMs on MLE-Smith tasks is strongly correlated with their performance on carefully human-designed tasks, highlighting the effectiveness of the MLE-Smith to scaling up MLE tasks, while maintaining task quality.

---

### Cocoon: A System Architecture for Differentially Private Training with Correlated Noises

Authors: Donghwan Kim, Xin Gu, Jinho Baek, Timothy Lo, Younghoon Min, Kwangsik Shin, Jongryool Kim, Jongse Park, Kiwan Maeng
Published: 2025-10-08
Categories: cs.AR, cs.AI, cs.CR, cs.LG
arXiv: https://arxiv.org/abs/2510.07304v2

Machine learning (ML) models memorize and leak training data, causing serious privacy issues to data owners. Training algorithms with differential privacy (DP) have been gaining attention as a solution. However, these algorithms add noise at each training iteration and degrade accuracy, limiting their real-world adoption. To improve accuracy, a new family of approaches adds carefully designed correlated noises, so that noises cancel out each other across iterations. We performed an extensive characterization study of these new mechanisms and show they incur non-negligible overheads when the model is relatively large or uses large embedding tables compared to the hardware capacity. Motivated by the analysis, we propose Cocoon, a framework for efficient training with correlated noises. Cocoon stores and processes the large noise history across CPU, GPU, and memory extension module, introduces optimizations for sparse embedding tables, and leverages to-be-commercialized near-memory processing (NMP) devices. On a real system with an FPGA-based NMP device prototype, Cocoon improves the performance by 1.23-10.82x.

---

### SpecGuard: Spectral Projection-based Advanced Invisible Watermarking

Authors: Inzamamul Alam, Md Tanvir Islam, Khan Muhammad, Simon S. Woo
Published: 2025-10-08
Categories: cs.CV
arXiv: https://arxiv.org/abs/2510.07302v1

Watermarking embeds imperceptible patterns into images for authenticity verification. However, existing methods often lack robustness against various transformations primarily including distortions, image regeneration, and adversarial perturbation, creating real-world challenges. In this work, we introduce SpecGuard, a novel watermarking approach for robust and invisible image watermarking. Unlike prior approaches, we embed the message inside hidden convolution layers by converting from the spatial domain to the frequency domain using spectral projection of a higher frequency band that is decomposed by wavelet projection. Spectral projection employs Fast Fourier Transform approximation to transform spatial data into the frequency domain efficiently. In the encoding phase, a strength factor enhances resilience against diverse attacks, including adversarial, geometric, and regeneration-based distortions, ensuring the preservation of copyrighted information. Meanwhile, the decoder leverages Parseval's theorem to effectively learn and extract the watermark pattern, enabling accurate retrieval under challenging transformations. We evaluate the proposed SpecGuard based on the embedded watermark's invisibility, capacity, and robustness. Comprehensive experiments demonstrate the proposed SpecGuard outperforms the state-of-the-art models. To ensure reproducibility, the full code is released on \href{https://github.com/inzamamulDU/SpecGuard_ICCV_2025}{\textcolor{blue}{\textbf{GitHub}}}.

---

### Think Natively: Unlocking Multilingual Reasoning with Consistency-Enhanced Reinforcement Learning

Authors: Xue Zhang, Yunlong Liang, Fandong Meng, Songming Zhang, Kaiyu Huang, Yufeng Chen, Jinan Xu, Jie Zhou
Published: 2025-10-08
Categories: cs.CL
arXiv: https://arxiv.org/abs/2510.07300v3

Large Reasoning Models (LRMs) have achieved remarkable performance on complex reasoning tasks by adopting the ``think-then-answer'' paradigm, which enhances both accuracy and interpretability. However, current LRMs exhibit two critical limitations when processing non-English languages: (1) They often struggle to maintain input-output language consistency; (2) They generally perform poorly with wrong reasoning paths and lower answer accuracy compared to English. These limitations significantly compromise the interpretability of reasoning processes and degrade the user experience for non-English speakers, hindering the global deployment of LRMs. To address these limitations, we propose M-Thinker, which is trained by the GRPO algorithm that involves a Language Consistency (LC) reward and a novel Cross-lingual Thinking Alignment (CTA) reward. Specifically, the LC reward defines a strict constraint on the language consistency between the input, thought, and answer. Besides, the CTA reward compares the model's non-English reasoning paths with its English reasoning path to transfer its own reasoning capability from English to non-English languages. Through an iterative RL procedure, our M-Thinker-1.5B/4B/7B models not only achieve nearly 100% language consistency and superior performance on two multilingual benchmarks (MMATH and PolyMath), but also exhibit excellent generalization on out-of-domain languages.

---

### Agentic generative AI for media content discovery at the national football league

Authors: Henry Wang, Md Sirajus Salekin, Jake Lee, Ross Claytor, Shinan Zhang, Michael Chi
Published: 2025-10-08
Categories: cs.AI
arXiv: https://arxiv.org/abs/2510.07297v1

Generative AI has unlocked new possibilities in content discovery and management. Through collaboration with the National Football League (NFL), we demonstrate how a generative-AI based workflow enables media researchers and analysts to query relevant historical plays using natural language rather than traditional filter-and-click interfaces. The agentic workflow takes a user query as input, breaks it into elements, and translates them into the underlying database query language. Accuracy and latency are further improved through carefully designed semantic caching. The solution achieves over 95 percent accuracy and reduces the average time to find relevant videos from 10 minutes to 30 seconds, significantly increasing the NFL's operational efficiency and allowing users to focus on producing creative content and engaging storylines.

---

### AudioMarathon: A Comprehensive Benchmark for Long-Context Audio Understanding and Efficiency in Audio LLMs

Authors: Peize He, Zichen Wen, Yubo Wang, Yuxuan Wang, Xiaoqian Liu, Jiajie Huang, Zehui Lei, Zhuangcheng Gu, Xiangqi Jin, Jiabing Yang, Kai Li, Zhifei Liu et al.
Published: 2025-10-08
Categories: cs.SD, cs.AI, cs.CL, eess.AS
arXiv: https://arxiv.org/abs/2510.07293v1

Processing long-form audio is a major challenge for Large Audio Language models (LALMs). These models struggle with the quadratic cost of attention ($O(N^2)$) and with modeling long-range temporal dependencies. Existing audio benchmarks are built mostly from short clips and do not evaluate models in realistic long context settings. To address this gap, we introduce AudioMarathon, a benchmark designed to evaluate both understanding and inference efficiency on long-form audio. AudioMarathon provides a diverse set of tasks built upon three pillars: long-context audio inputs with durations ranging from 90.0 to 300.0 seconds, which correspond to encoded sequences of 2,250 to 7,500 audio tokens, respectively, full domain coverage across speech, sound, and music, and complex reasoning that requires multi-hop inference. We evaluate state-of-the-art LALMs and observe clear performance drops as audio length grows. We also study acceleration techniques and analyze the trade-offs of token pruning and KV cache eviction. The results show large gaps across current LALMs and highlight the need for better temporal reasoning and memory-efficient architectures. We believe AudioMarathon will drive the audio and multimodal research community to develop more advanced audio understanding models capable of solving complex audio tasks.

---

### On the Convergence of Moral Self-Correction in Large Language Models

Authors: Guangliang Liu, Haitao Mao, Bochuan Cao, Zhiyu Xue, Xitong Zhang, Rongrong Wang, Kristen Marie Johnson
Published: 2025-10-08
Categories: cs.CL, cs.LG
arXiv: https://arxiv.org/abs/2510.07290v3

Large Language Models (LLMs) are able to improve their responses when instructed to do so, a capability known as self-correction. When instructions provide only a general and abstract goal without specific details about potential issues in the response, LLMs must rely on their internal knowledge to improve response quality, a process referred to as intrinsic self-correction. The empirical success of intrinsic self-correction is evident in various applications, but how and why it is effective remains unknown. Focusing on moral self-correction in LLMs, we reveal a key characteristic of intrinsic self-correction: performance convergence through multi-round interactions; and provide a mechanistic analysis of this convergence behavior. Based on our experimental results and analysis, we uncover the underlying mechanism of convergence: consistently injected self-correction instructions activate moral concepts that reduce model uncertainty, leading to converged performance as the activated moral concepts stabilize over successive rounds. This paper demonstrates the strong potential of moral self-correction by showing that it exhibits a desirable property of converged performance.

---

### MolGA: Molecular Graph Adaptation with Pre-trained 2D Graph Encoder

Authors: Xingtong Yu, Chang Zhou, Xinming Zhang, Yuan Fang
Published: 2025-10-08
Categories: cs.LG
arXiv: https://arxiv.org/abs/2510.07289v3

Molecular graph representation learning is widely used in chemical and biomedical research. While pre-trained 2D graph encoders have demonstrated strong performance, they overlook the rich molecular domain knowledge associated with submolecular instances (atoms and bonds). While molecular pre-training approaches incorporate such knowledge into their pre-training objectives, they typically employ designs tailored to a specific type of knowledge, lacking the flexibility to integrate diverse knowledge present in molecules. Hence, reusing widely available and well-validated pre-trained 2D encoders, while incorporating molecular domain knowledge during downstream adaptation, offers a more practical alternative. In this work, we propose MolGA, which adapts pre-trained 2D graph encoders to downstream molecular applications by flexibly incorporating diverse molecular domain knowledge. First, we propose a molecular alignment strategy that bridge the gap between pre-trained topological representations with domain-knowledge representations. Second, we introduce a conditional adaptation mechanism that generates instance-specific tokens to enable fine-grained integration of molecular domain knowledge for downstream tasks. Finally, we conduct extensive experiments on eleven public datasets, demonstrating the effectiveness of MolGA.

---

### Evolutionary Profiles for Protein Fitness Prediction

Authors: Jigang Fan, Xiaoran Jiao, Shengdong Lin, Zhanming Liang, Weian Mao, Chenchen Jing, Hao Chen, Chunhua Shen
Published: 2025-10-08
Categories: cs.LG, cs.AI, q-bio.BM, q-bio.QM
arXiv: https://arxiv.org/abs/2510.07286v3

Predicting the fitness impact of mutations is central to protein engineering but constrained by limited assays relative to the size of sequence space. Protein language models (pLMs) trained with masked language modeling (MLM) exhibit strong zero-shot fitness prediction; we provide a unifying view by interpreting natural evolution as implicit reward maximization and MLM as inverse reinforcement learning (IRL), in which extant sequences act as expert demonstrations and pLM log-odds serve as fitness estimates. Building on this perspective, we introduce EvoIF, a lightweight model that integrates two complementary sources of evolutionary signal: (i) within-family profiles from retrieved homologs and (ii) cross-family structural-evolutionary constraints distilled from inverse folding logits. EvoIF fuses sequence-structure representations with these profiles via a compact transition block, yielding calibrated probabilities for log-odds scoring. On ProteinGym (217 mutational assays; >2.5M mutants), EvoIF and its MSA-enabled variant achieve state-of-the-art or competitive performance while using only 0.15% of the training data and fewer parameters than recent large models. Ablations confirm that within-family and cross-family profiles are complementary, improving robustness across function types, MSA depths, taxa, and mutation depths. The codes will be made publicly available.

---

### GTCN-G: A Residual Graph-Temporal Fusion Network for Imbalanced Intrusion Detection

Authors: Tianxiang Xu, Zhichao Wen, Xinyu Zhao, Qi Hu, Yan Li, Chang Liu
Published: 2025-10-08
Categories: cs.LG, cs.AI
arXiv: https://arxiv.org/abs/2510.07285v3

The escalating complexity of network threats and the inherent class imbalance in traffic data present formidable challenges for modern Intrusion Detection Systems (IDS). While Graph Neural Networks (GNNs) excel in modeling topological structures and Temporal Convolutional Networks (TCNs) are proficient in capturing time-series dependencies, a framework that synergistically integrates both while explicitly addressing data imbalance remains an open challenge. This paper introduces a novel deep learning framework, named Gated Temporal Convolutional Network and Graph (GTCN-G), engineered to overcome these limitations. Our model uniquely fuses a Gated TCN (G-TCN) for extracting hierarchical temporal features from network flows with a Graph Convolutional Network (GCN) designed to learn from the underlying graph structure. The core innovation lies in the integration of a residual learning mechanism, implemented via a Graph Attention Network (GAT). This mechanism preserves original feature information through residual connections, which is critical for mitigating the class imbalance problem and enhancing detection sensitivity for rare malicious activities (minority classes). We conducted extensive experiments on two public benchmark datasets, UNSW-NB15 and ToN-IoT, to validate our approach. The empirical results demonstrate that the proposed GTCN-G model achieves state-of-the-art performance, significantly outperforming existing baseline models in both binary and multi-class classification tasks.

---

### Online Rubrics Elicitation from Pairwise Comparisons

Authors: MohammadHossein Rezaei, Robert Vacareanu, Zihao Wang, Clinton Wang, Bing Liu, Yunzhong He, Afra Feyza Akyürek
Published: 2025-10-08
Categories: cs.CL, cs.AI, cs.LG
arXiv: https://arxiv.org/abs/2510.07284v2

Rubrics provide a flexible way to train LLMs on open-ended long-form answers where verifiable rewards are not applicable and human preferences provide coarse signals. Prior work shows that reinforcement learning with rubric-based rewards leads to consistent gains in LLM post-training. Most existing approaches rely on rubrics that remain static over the course of training. Such static rubrics, however, are vulnerable to reward-hacking type behaviors and fail to capture emergent desiderata that arise during training. We introduce Online Rubrics Elicitation (OnlineRubrics), a method that dynamically curates evaluation criteria in an online manner through pairwise comparisons of responses from current and reference policies. This online process enables continuous identification and mitigation of errors as training proceeds. Empirically, this approach yields consistent improvements of up to 8% over training exclusively with static rubrics across AlpacaEval, GPQA, ArenaHard as well as the validation sets of expert questions and rubrics. We qualitatively analyze the elicited criteria and identify prominent themes such as transparency, practicality, organization, and reasoning.

---

### Evaluating Fundus-Specific Foundation Models for Diabetic Macular Edema Detection

Authors: Franco Javier Arellano, José Ignacio Orlando
Published: 2025-10-08
Categories: cs.CV
arXiv: https://arxiv.org/abs/2510.07277v2

Diabetic Macular Edema (DME) is a leading cause of vision loss among patients with Diabetic Retinopathy (DR). While deep learning has shown promising results for automatically detecting this condition from fundus images, its application remains challenging due the limited availability of annotated data. Foundation Models (FM) have emerged as an alternative solution. However, it is unclear if they can cope with DME detection in particular. In this paper, we systematically compare different FM and standard transfer learning approaches for this task. Specifically, we compare the two most popular FM for retinal images-RETFound and FLAIR-and an EfficientNetB0 backbone, across different training regimes and evaluation settings in IDRiD, MESSIDOR-2 and OCT-and-Eye-FundusImages (OEFI). Results show that despite their scale, FM do not consistently outperform fine-tuned CNNs in this task. In particular, EfficientNet-B0 consistently achieves competitive or superior performance across evaluation settings, with FLAIR being the most competitive foundation model, consistently outperforming RETFound. These findings suggest that FMs do not necessarily provide an advantage for fine-grained ophthalmic tasks such as DME detection, even after fine-tuning, highlighting lightweight CNNs as strong baselines in data-scarce environments.

---

### Multi-Objective Multi-Agent Path Finding with Lexicographic Cost Preferences

Authors: Pulkit Rustagi, Kyle Hollins Wray, Sandhya Saisubramanian
Published: 2025-10-08
Categories: cs.AI, cs.MA
arXiv: https://arxiv.org/abs/2510.07276v2

Many real-world scenarios require multiple agents to coordinate in shared environments, while balancing trade-offs between multiple, potentially competing objectives. Current multi-objective multi-agent path finding (MO-MAPF) algorithms typically produce conflict-free plans by computing Pareto frontiers. They do not explicitly optimize for user-defined preferences, even when the preferences are available, and scale poorly with the number of objectives. We propose a lexicographic framework for modeling MO-MAPF, along with an algorithm \textit{Lexicographic Conflict-Based Search} (LCBS) that directly computes a single solution aligned with a lexicographic preference over objectives. LCBS integrates a priority-aware low-level $A^*$ search with conflict-based search, avoiding Pareto frontier construction and enabling efficient planning guided by preference over objectives. We provide insights into optimality and scalability, and empirically demonstrate that LCBS computes optimal solutions while scaling to instances with up to ten objectives -- far beyond the limits of existing MO-MAPF methods. Evaluations on standard and randomized MAPF benchmarks show consistently higher success rates against state-of-the-art baselines, especially with increasing number of objectives.

---

### On the false election between regulation and innovation. Ideas for regulation through the responsible use of artificial intelligence in research and education.(Spanish version)

Authors: Pompeu Casanovas
Published: 2025-10-08
Categories: cs.CY, cs.AI
arXiv: https://arxiv.org/abs/2510.07268v1

This short essay is a reworking of the answers offered by the author at the Debate Session of the AIHUB (CSIC) and EduCaixa Summer School, organized by Marta Garcia-Matos and Lissette Lemus, and coordinated by Albert Sabater (OEIAC, UG), with the participation of Vanina Martinez-Posse (IIIA-CSIC), Eulalia Soler (Eurecat) and Pompeu Casanovas (IIIA-CSIC) on July 4th 2025. Albert Sabater posed three questions: (1) How can regulatory frameworks priori-tise the protection of fundamental rights (privacy, non-discrimination, autonomy, etc.) in the development of AI, without falling into the false dichotomy between regulation and innova-tion? (2) Given the risks of AI (bias, mass surveillance, manipulation), what examples of regu-lations or policies have demonstrated that it is possible to foster responsible innovation, putting the public interest before profitability, without giving in to competitive pressure from actors such as China or the US? (3) In a scenario where the US prioritizes flexibility, what mecha-nisms could ensure that international cooperation in AI does not become a race to the bottom in rights, but rather a global standard of accountability? The article attempts to answer these three questions and concludes with some reflections on the relevance of the answers for education and research.

---

### Dynamic Regret Bounds for Online Omniprediction with Long Term Constraints

Authors: Yahav Bechavod, Jiuyao Lu, Aaron Roth
Published: 2025-10-08
Categories: cs.LG, cs.GT
arXiv: https://arxiv.org/abs/2510.07266v1

We present an algorithm guaranteeing dynamic regret bounds for online omniprediction with long term constraints. The goal in this recently introduced problem is for a learner to generate a sequence of predictions which are broadcast to a collection of downstream decision makers. Each decision maker has their own utility function, as well as a vector of constraint functions, each mapping their actions and an adversarially selected state to reward or constraint violation terms. The downstream decision makers select actions "as if" the state predictions are correct, and the goal of the learner is to produce predictions such that all downstream decision makers choose actions that give them worst-case utility guarantees while minimizing worst-case constraint violation. Within this framework, we give the first algorithm that obtains simultaneous \emph{dynamic regret} guarantees for all of the agents -- where regret for each agent is measured against a potentially changing sequence of actions across rounds of interaction, while also ensuring vanishing constraint violation for each agent. Our results do not require the agents themselves to maintain any state -- they only solve one-round constrained optimization problems defined by the prediction made at that round.

---

### Test-Time Graph Search for Goal-Conditioned Reinforcement Learning

Authors: Evgenii Opryshko, Junwei Quan, Claas Voelcker, Yilun Du, Igor Gilitschenski
Published: 2025-10-08
Categories: cs.LG
arXiv: https://arxiv.org/abs/2510.07257v2

Offline goal-conditioned reinforcement learning (GCRL) often struggles with long-horizon tasks, where errors in value estimation accumulate and produce unreliable policies. It is typically assumed that effective long-term planning is infeasible without specialized training. In contrast, our work demonstrates that existing GCRL policies can complete long-horizon tasks when combined with a lightweight, training-free planning wrapper. We find that standard goal-conditioned value functions encode locally consistent geometric structure sufficient for planning. Our approach, Test-Time Graph Search (TTGS), constructs a graph over the offline dataset and employs an adaptive subgoal selection strategy. To address unreliable value estimates during shortest-path search, we propose a novel mechanism that softly penalizes long-distance transitions. Our method incurs negligible computational overhead and requires no additional supervision or parameter updates. On the OGBench benchmark, TTGS significantly boosts success rates across multiple base learners and tasks, with primary gains on challenging long-horizon locomotion tasks where some success rates are improved from near-zero to over 90\%, often matching or outperforming methods that require complex auxiliary training. Code and videos can be found at https://ktolnos.github.io/ttgs.

---

### Efficient reductions from a Gaussian source with applications to statistical-computational tradeoffs

Authors: Mengqi Lou, Guy Bresler, Ashwin Pananjady
Published: 2025-10-08
Categories: math.ST, cs.CC, cs.IT, stat.ML
arXiv: https://arxiv.org/abs/2510.07250v1

Given a single observation from a Gaussian distribution with unknown mean $θ$, we design computationally efficient procedures that can approximately generate an observation from a different target distribution $Q_θ$ uniformly for all $θ$ in a parameter set. We leverage our technique to establish reduction-based computational lower bounds for several canonical high-dimensional statistical models under widely-believed conjectures in average-case complexity. In particular, we cover cases in which: 1. $Q_θ$ is a general location model with non-Gaussian distribution, including both light-tailed examples (e.g., generalized normal distributions) and heavy-tailed ones (e.g., Student's $t$-distributions). As a consequence, we show that computational lower bounds proved for spiked tensor PCA with Gaussian noise are universal, in that they extend to other non-Gaussian noise distributions within our class. 2. $Q_θ$ is a normal distribution with mean $f(θ)$ for a general, smooth, and nonlinear link function $f:\mathbb{R} \rightarrow \mathbb{R}$. Using this reduction, we construct a reduction from symmetric mixtures of linear regressions to generalized linear models with link function $f$, and establish computational lower bounds for solving the $k$-sparse generalized linear model when $f$ is an even function. This result constitutes the first reduction-based confirmation of a $k$-to-$k^2$ statistical-to-computational gap in $k$-sparse phase retrieval, resolving a conjecture posed by Cai et al. (2016). As a second application, we construct a reduction from the sparse rank-1 submatrix model to the planted submatrix model, establishing a pointwise correspondence between the phase diagrams of the two models that faithfully preserves regions of computational hardness and tractability.

---

### TalkCuts: A Large-Scale Dataset for Multi-Shot Human Speech Video Generation

Authors: Jiaben Chen, Zixin Wang, Ailing Zeng, Yang Fu, Xueyang Yu, Siyuan Cen, Julian Tanke, Yihang Chen, Koichi Saito, Yuki Mitsufuji, Chuang Gan
Published: 2025-10-08
Categories: cs.CV
arXiv: https://arxiv.org/abs/2510.07249v2

In this work, we present TalkCuts, a large-scale dataset designed to facilitate the study of multi-shot human speech video generation. Unlike existing datasets that focus on single-shot, static viewpoints, TalkCuts offers 164k clips totaling over 500 hours of high-quality human speech videos with diverse camera shots, including close-up, half-body, and full-body views. The dataset includes detailed textual descriptions, 2D keypoints and 3D SMPL-X motion annotations, covering over 10k identities, enabling multimodal learning and evaluation. As a first attempt to showcase the value of the dataset, we present Orator, an LLM-guided multi-modal generation framework as a simple baseline, where the language model functions as a multi-faceted director, orchestrating detailed specifications for camera transitions, speaker gesticulations, and vocal modulation. This architecture enables the synthesis of coherent long-form videos through our integrated multi-modal video generation module. Extensive experiments in both pose-guided and audio-driven settings show that training on TalkCuts significantly enhances the cinematographic coherence and visual appeal of generated multi-shot speech videos. We believe TalkCuts provides a strong foundation for future work in controllable, multi-shot speech video generation and broader multimodal learning.

---

### Don't Adapt Small Language Models for Tools; Adapt Tool Schemas to the Models

Authors: Jonggeun Lee, Woojung Song, Jongwook Han, Haesung Pyun, Yohan Jo
Published: 2025-10-08
Categories: cs.CL
arXiv: https://arxiv.org/abs/2510.07248v3

Small language models (SLMs) enable scalable tool-augmented multi-agent systems where multiple SLMs handle subtasks orchestrated by a powerful coordinator. However, they struggle with tool-use tasks, particularly in selecting appropriate tools and identifying correct parameters. A common failure mode is \textit{schema misalignment}: models hallucinate plausible tool names that are absent from the provided tool schema, due to different naming conventions internalized during pretraining. Rather than training models to adapt to unfamiliar schemas, we propose adapting schemas to align with models' pretrained knowledge. We introduce \textbf{PA-Tool} (Pretraining-Aligned Tool Schema Generation), a training-free method that leverages peakedness, a signal used in contamination detection that indicates pretraining familiarity, to rename tool components. By generating multiple candidates and selecting the candidate with the highest peakedness, PA-Tool identifies pretraining-aligned naming patterns. Experiments on MetaTool and RoTBench show improvements of up to 17\%, with schema misalignment errors reduced by 80\%. PA-Tool enables small models to substantially improve tool-use accuracy without retraining, showing that schema-level interventions can unlock the tool-use potential of resource-efficient models. Our code is available at https://github.com/holi-lab/PA-Tool.

---

### Discriminative Feature Feedback with General Teacher Classes

Authors: Omri Bar Oz, Tosca Lechner, Sivan Sabato
Published: 2025-10-08
Categories: cs.LG
arXiv: https://arxiv.org/abs/2510.07245v2

We study the theoretical properties of the interactive learning protocol Discriminative Feature Feedback (DFF) (Dasgupta et al., 2018). The DFF learning protocol uses feedback in the form of discriminative feature explanations. We provide the first systematic study of DFF in a general framework that is comparable to that of classical protocols such as supervised learning and online learning. We study the optimal mistake bound of DFF in the realizable and the non-realizable settings, and obtain novel structural results, as well as insights into the differences between Online Learning and settings with richer feedback such as DFF. We characterize the mistake bound in the realizable setting using a new notion of dimension. In the non-realizable setting, we provide a mistake upper bound and show that it cannot be improved in general. Our results show that unlike Online Learning, in DFF the realizable dimension is insufficient to characterize the optimal non-realizable mistake bound or the existence of no-regret algorithms.

---

### LeMAJ (Legal LLM-as-a-Judge): Bridging Legal Reasoning and LLM Evaluation

Authors: Joseph Enguehard, Morgane Van Ermengem, Kate Atkinson, Sujeong Cha, Arijit Ghosh Chowdhury, Prashanth Kallur Ramaswamy, Jeremy Roghair, Hannah R Marlowe, Carina Suzana Negreanu, Kitty Boxall, Diana Mincu
Published: 2025-10-08
Categories: cs.CL, cs.AI
arXiv: https://arxiv.org/abs/2510.07243v1

Evaluating large language model (LLM) outputs in the legal domain presents unique challenges due to the complex and nuanced nature of legal analysis. Current evaluation approaches either depend on reference data, which is costly to produce, or use standardized assessment methods, both of which have significant limitations for legal applications. Although LLM-as-a-Judge has emerged as a promising evaluation technique, its reliability and effectiveness in legal contexts depend heavily on evaluation processes unique to the legal industry and how trustworthy the evaluation appears to the human legal expert. This is where existing evaluation methods currently fail and exhibit considerable variability. This paper aims to close the gap: a) we break down lengthy responses into 'Legal Data Points' (LDPs), self-contained units of information, and introduce a novel, reference-free evaluation methodology that reflects how lawyers evaluate legal answers; b) we demonstrate that our method outperforms a variety of baselines on both our proprietary dataset and an open-source dataset (LegalBench); c) we show how our method correlates more closely with human expert evaluations and helps improve inter-annotator agreement; and finally d) we open source our Legal Data Points for a subset of LegalBench used in our experiments, allowing the research community to replicate our results and advance research in this vital area of LLM evaluation on legal question-answering.

---

### Hybrid Reinforcement: When Reward Is Sparse, It's Better to Be Dense

Authors: Leitian Tao, Ilia Kulikov, Swarnadeep Saha, Tianlu Wang, Jing Xu, Sharon Li, Jason E Weston, Ping Yu
Published: 2025-10-08
Categories: cs.CL, cs.LG
arXiv: https://arxiv.org/abs/2510.07242v3

Post-training for reasoning of large language models (LLMs) increasingly relies on verifiable rewards: deterministic checkers that provide 0-1 correctness signals. While reliable, such binary feedback is brittle--many tasks admit partially correct or alternative answers that verifiers under-credit, and the resulting all-or-nothing supervision limits learning. Reward models offer richer, continuous feedback, which can serve as a complementary supervisory signal to verifiers. We introduce HERO (Hybrid Ensemble Reward Optimization), a reinforcement learning framework that integrates verifier signals with reward-model scores in a structured way. HERO employs stratified normalization to bound reward-model scores within verifier-defined groups, preserving correctness while refining quality distinctions, and variance-aware weighting to emphasize challenging prompts where dense signals matter most. Across diverse mathematical reasoning benchmarks, HERO consistently outperforms RM-only and verifier-only baselines, with strong gains on both verifiable and hard-to-verify tasks. Our results show that hybrid reward design retains the stability of verifiers while leveraging the nuance of reward models to advance reasoning.

---

### Red-Bandit: Test-Time Adaptation for LLM Red-Teaming via Bandit-Guided LoRA Experts

Authors: Christos Ziakas, Nicholas Loo, Nishita Jain, Alessandra Russo
Published: 2025-10-08
Categories: cs.CL
arXiv: https://arxiv.org/abs/2510.07239v2

Automated red-teaming has emerged as a scalable approach for auditing Large Language Models (LLMs) prior to deployment, yet existing approaches lack mechanisms to efficiently adapt to model-specific vulnerabilities at inference. We introduce Red-Bandit, a red-teaming framework that adapts online to identify and exploit model failure modes under distinct attack styles (e.g., manipulation, slang). Red-Bandit post-trains a set of parameter-efficient LoRA experts, each specialized for a particular attack style, using reinforcement learning that rewards the generation of unsafe prompts via a rule-based safety model. At inference, a multi-armed bandit policy dynamically selects among these attack-style experts based on the target model's response safety, balancing exploration and exploitation. Red-Bandit achieves state-of-the-art results on AdvBench under sufficient exploration (ASR@10), while producing more human-readable prompts (lower perplexity). Moreover, Red-Bandit's bandit policy serves as a diagnostic tool for uncovering model-specific vulnerabilities by indicating which attack styles most effectively elicit unsafe behaviors.

---

### When Benchmarks Age: Temporal Misalignment through Large Language Model Factuality Evaluation

Authors: Xunyi Jiang, Dingyi Chang, Julian McAuley, Xin Xu
Published: 2025-10-08
Categories: cs.CL
arXiv: https://arxiv.org/abs/2510.07238v2

The rapid evolution of large language models (LLMs) and the real world has outpaced the static nature of widely used evaluation benchmarks, raising concerns about their reliability for evaluating LLM factuality. While substantial works continue to rely on the popular but old benchmarks, their temporal misalignment with real-world facts and modern LLMs, and their effects on LLM factuality evaluation remain underexplored. Therefore, in this work, we present a systematic investigation of this issue by examining five popular factuality benchmarks and eight LLMs released across different years. An up-to-date fact retrieval pipeline and three metrics are tailored to quantify benchmark aging and its impact on LLM factuality evaluation. Experimental results and analysis illustrate that a considerable portion of samples in the widely used factuality benchmarks are outdated, leading to unreliable assessments of LLM factuality. We hope our work can provide a testbed to assess the reliability of a benchmark for LLM factuality evaluation and inspire more research on the benchmark aging issue. Codes are available in https://github.com/JiangXunyi/BenchAge.

---

### LAD-RAG: Layout-aware Dynamic RAG for Visually-Rich Document Understanding

Authors: Zhivar Sourati, Zheng Wang, Marianne Menglin Liu, Yazhe Hu, Mengqing Guo, Sujeeth Bharadwaj, Kyu Han, Tao Sheng, Sujith Ravi, Morteza Dehghani, Dan Roth
Published: 2025-10-08
Categories: cs.CL
arXiv: https://arxiv.org/abs/2510.07233v2

Question answering over visually rich documents (VRDs) requires reasoning not only over isolated content but also over documents' structural organization and cross-page dependencies. However, conventional retrieval-augmented generation (RAG) methods encode content in isolated chunks during ingestion, losing structural and cross-page dependencies, and retrieve a fixed number of pages at inference, regardless of the specific demands of the question or context. This often results in incomplete evidence retrieval and degraded answer quality for multi-page reasoning tasks. To address these limitations, we propose LAD-RAG, a novel Layout-Aware Dynamic RAG framework. During ingestion, LAD-RAG constructs a symbolic document graph that captures layout structure and cross-page dependencies, adding it alongside standard neural embeddings to yield a more holistic representation of the document. During inference, an LLM agent dynamically interacts with the neural and symbolic indices to adaptively retrieve the necessary evidence based on the query. Experiments on MMLongBench-Doc, LongDocURL, DUDE, and MP-DocVQA demonstrate that LAD-RAG improves retrieval, achieving over 90% perfect recall on average without any top-k tuning, and outperforming baseline retrievers by up to 20% in recall at comparable noise levels, yielding higher QA accuracy with minimal latency.

---

### EconCausal: A Context-Aware Economic Reasoning Benchmark for Large Language Models

Authors: Donggyu Lee, Hyeok Yun, Meeyoung Cha, Sungwon Park, Sangyoon Park, Jihee Kim
Published: 2025-10-08
Categories: cs.CL, cs.AI
arXiv: https://arxiv.org/abs/2510.07231v5

Socio-economic causal effects depend heavily on their institutional and environmental contexts. The same intervention can produce different, even opposite, effects across regulatory regimes, market conditions, time periods, or populations. This poses a challenge for large language models (LLMs) in decision-support roles: can they infer the direction of a causal effect under a specified context, and revise that judgment when the context changes? To address this, we introduce EconCausal, a large-scale benchmark of 10,490 context-annotated causal triplets extracted from 2,595 high-quality empirical studies in top-tier economics and finance journals, constructed through a rigorous four-stage pipeline with multi-run consensus, context refinement, and multi-critic filtering. Across 19 models, we observe source-sign over-transfer in cross-context prediction. While top models reach ~84% accuracy in single-context prediction, mean accuracy is 74.75% overall but drops to 44.27% when source and target signs disagree. On these cases, accuracy is 11.54 percentage points (pp) lower with the signed source example than with the target input alone. Mean accuracy is below 60% under corrupted source signs. Models also struggle to predict study-reported nonsignificant effects (19.16% accuracy). The dataset and benchmark are publicly available at https://anonymous.4open.science/r/econcausal-benchmark-6F12.

---

### Customer-R1: Personalized Simulation of Human Behaviors via RL-based LLM Agent in Online Shopping

Authors: Ziyi Wang, Yuxuan Lu, Yimeng Zhang, Jing Huang, Dakuo Wang
Published: 2025-10-08
Categories: cs.CL
arXiv: https://arxiv.org/abs/2510.07230v2

Simulating step-wise human behavior with Large Language Models (LLMs) has become an emerging research direction, enabling applications in various practical domains. While prior methods, including prompting, supervised fine-tuning (SFT), and reinforcement learning (RL), have shown promise in modeling step-wise behavior, they primarily learn a population-level policy without conditioning on a user's persona, yielding generic rather than personalized simulations. In this work, we pose a critical question: how can LLM agents better simulate personalized user behavior? We introduce Customer-R1, an RL-based method for personalized, step-wise user behavior simulation in online shopping environments. Our policy is conditioned on an explicit persona, and we optimize next-step rationale and action generation via action correctness reward signals. Experiments on the OPeRA dataset emonstrate that Customer-R1 not only significantly outperforms prompting and SFT-based baselines in next-action prediction tasks, but also better matches users' action distribution, indicating higher fidelity in personalized behavior simulation.

---

### Where to Begin: Efficient Pretraining via Subnetwork Selection and Distillation

Authors: Arjun Krishnakumar, Rhea Sanjay Sukthanker, Hannan Javed Mahadik, Gabriela Kadlecová, Vladyslav Moroshan, Timur Carstensen, Frank Hutter, Aaron Klein
Published: 2025-10-08
Categories: cs.CL, cs.AI
arXiv: https://arxiv.org/abs/2510.07227v2

Small Language models (SLMs) offer an efficient and accessible alternative to Large Language Models (LLMs), delivering strong performance while using far fewer resources. We introduce a simple and effective framework for pretraining SLMs that brings together three complementary ideas. First, we identify structurally sparse sub-network initializations that consistently outperform randomly initialized models of similar size under the same compute budget. Second, we use evolutionary search to automatically discover high-quality sub-network initializations, providing better starting points for pretraining. Third, we apply knowledge distillation from larger teacher models to speed up training and improve generalization. Together, these components make SLM pretraining substantially more efficient: our best model, discovered using evolutionary search and initialized with LLM weights, matches the validation perplexity of a comparable Pythia SLM while requiring 5.16x and 1.26x fewer floating point operations for token budgets of 10B and 100B, respectively. We release all code publicly, offering a practical and reproducible path toward cost-efficient small language model development at scale.

---

### Machines in the Crowd? Measuring the Footprint of Machine-Generated Text on Reddit

Authors: Lucio La Cava, Luca Maria Aiello, Andrea Tagarelli
Published: 2025-10-08
Categories: cs.SI, cs.CL, cs.CY, physics.soc-ph
arXiv: https://arxiv.org/abs/2510.07226v1

Generative Artificial Intelligence is reshaping online communication by enabling large-scale production of Machine-Generated Text (MGT) at low cost. While its presence is rapidly growing across the Web, little is known about how MGT integrates into social media environments. In this paper, we present the first large-scale characterization of MGT on Reddit. Using a state-of-the-art statistical method for detection of MGT, we analyze over two years of activity (2022-2024) across 51 subreddits representative of Reddit's main community types such as information seeking, social support, and discussion. We study the concentration of MGT across communities and over time, and compared MGT to human-authored text in terms of social signals it expresses and engagement it receives. Our very conservative estimate of MGT prevalence indicates that synthetic text is marginally present on Reddit, but it can reach peaks of up to 9% in some communities in some months. MGT is unevenly distributed across communities, more prevalent in subreddits focused on technical knowledge and social support, and often concentrated in the activity of a small fraction of users. MGT also conveys distinct social signals of warmth and status giving typical of language of AI assistants. Despite these stylistic differences, MGT achieves engagement levels comparable than human-authored content and in a few cases even higher, suggesting that AI-generated text is becoming an organic component of online social discourse. This work offers the first perspective on the MGT footprint on Reddit, paving the way for new investigations involving platform governance, detection strategies, and community dynamics.

---

### How much speech data is necessary for ASR in African languages? An evaluation of data scaling in Kinyarwanda and Kikuyu

Authors: Benjamin Akera, Evelyn Nafula, Patrick Walukagga, Gilbert Yiga, John Quinn, Ernest Mwebaze
Published: 2025-10-08
Categories: cs.CL
arXiv: https://arxiv.org/abs/2510.07221v1

The development of Automatic Speech Recognition (ASR) systems for low-resource African languages remains challenging due to limited transcribed speech data. While recent advances in large multilingual models like OpenAI's Whisper offer promising pathways for low-resource ASR development, critical questions persist regarding practical deployment requirements. This paper addresses two fundamental concerns for practitioners: determining the minimum data volumes needed for viable performance and characterizing the primary failure modes that emerge in production systems. We evaluate Whisper's performance through comprehensive experiments on two Bantu languages: systematic data scaling analysis on Kinyarwanda using training sets from 1 to 1,400 hours, and detailed error characterization on Kikuyu using 270 hours of training data. Our scaling experiments demonstrate that practical ASR performance (WER < 13\%) becomes achievable with as little as 50 hours of training data, with substantial improvements continuing through 200 hours (WER < 10\%). Complementing these volume-focused findings, our error analysis reveals that data quality issues, particularly noisy ground truth transcriptions, account for 38.6\% of high-error cases, indicating that careful data curation is as critical as data volume for robust system performance. These results provide actionable benchmarks and deployment guidance for teams developing ASR systems across similar low-resource language contexts. We release accompanying and models see https://github.com/SunbirdAI/kinyarwanda-whisper-eval

---

### GenPilot: A Multi-Agent System for Test-Time Prompt Optimization in Image Generation

Authors: Wen Ye, Zhaocheng Liu, Yuwei Gui, Tingyu Yuan, Yunyue Su, Bowen Fang, Chaoyang Zhao, Qiang Liu, Liang Wang
Published: 2025-10-08
Categories: cs.CV, cs.AI
arXiv: https://arxiv.org/abs/2510.07217v1

Text-to-image synthesis has made remarkable progress, yet accurately interpreting complex and lengthy prompts remains challenging, often resulting in semantic inconsistencies and missing details. Existing solutions, such as fine-tuning, are model-specific and require training, while prior automatic prompt optimization (APO) approaches typically lack systematic error analysis and refinement strategies, resulting in limited reliability and effectiveness. Meanwhile, test-time scaling methods operate on fixed prompts and on noise or sample numbers, limiting their interpretability and adaptability. To solve these, we introduce a flexible and efficient test-time prompt optimization strategy that operates directly on the input text. We propose a plug-and-play multi-agent system called GenPilot, integrating error analysis, clustering-based adaptive exploration, fine-grained verification, and a memory module for iterative optimization. Our approach is model-agnostic, interpretable, and well-suited for handling long and complex prompts. Simultaneously, we summarize the common patterns of errors and the refinement strategy, offering more experience and encouraging further exploration. Experiments on DPG-bench and Geneval with improvements of up to 16.9% and 5.7% demonstrate the strong capability of our methods in enhancing the text and image consistency and structural coherence of generated images, revealing the effectiveness of our test-time prompt optimization strategy. The code is available at https://github.com/27yw/GenPilot.

---

### Language Lives in Sparse Dimensions: Toward Interpretable and Efficient Multilingual Control for Large Language Models

Authors: Chengzhi Zhong, Fei Cheng, Qianying Liu, Yugo Murawaki, Chenhui Chu, Sadao Kurohashi
Published: 2025-10-08
Categories: cs.CL, cs.AI
arXiv: https://arxiv.org/abs/2510.07213v2

Large language models exhibit strong multilingual capabilities despite limited exposure to non-English data. Prior studies show that English-centric large language models map multilingual content into English-aligned representations at intermediate layers and then project them back into target-language token spaces in the final layer. From this observation, we hypothesize that this cross-lingual transition is governed by a small and sparse set of dimensions, which occur at consistent indices across the intermediate to final layers. Building on this insight, we introduce a simple, training-free method to identify and manipulate these dimensions, requiring only as few as 50 sentences of either parallel or monolingual data. Experiments on a multilingual generation control task reveal the interpretability of these dimensions, demonstrating that the interventions in these dimensions can switch the output language while preserving semantic content, and that it surpasses the performance of prior neuron-based approaches at a substantially lower cost.

---

### HyPlan: Hybrid Learning-Assisted Planning Under Uncertainty for Safe Autonomous Driving

Authors: Donald Pfaffmann, Matthias Klusch, Marcel Steinmetz
Published: 2025-10-08
Categories: cs.RO, cs.AI
arXiv: https://arxiv.org/abs/2510.07210v2

We present a novel hybrid learning-assisted planning method, named HyPlan, for solving the collision-free navigation problem for self-driving cars in partially observable traffic environments. HyPlan combines methods for multi-agent behavior prediction, deep reinforcement learning with proximal policy optimization and approximated online POMDP planning with heuristic confidence-based vertical pruning to reduce its execution time without compromising safety of driving. Our experimental performance analysis on the CARLA-CTS2 benchmark of critical traffic scenarios with pedestrians revealed that HyPlan may navigate safer than selected relevant baselines and perform significantly faster than considered alternative online POMDP planners.

---

### A Broader View of Thompson Sampling

Authors: Yanlin Qu, Hongseok Namkoong, Assaf Zeevi
Published: 2025-10-08
Categories: cs.LG
arXiv: https://arxiv.org/abs/2510.07208v2

Thompson Sampling is one of the most widely used and studied bandit algorithms, known for its simple structure, low regret performance, and solid theoretical guarantees. Yet, in stark contrast to most other families of bandit algorithms, the exact mechanism through which posterior sampling (as introduced by Thompson) is able to "properly" balance exploration and exploitation, remains a mystery. In this paper, we show that the core insight to address this question stems from recasting Thompson Sampling as an online optimization algorithm. To distill this, we introduce a suitable time invariant notion of regret that leads to a stationarized bandit problem, and a stationary Bellman-optimal policy. We then show that Thompson Sampling admits an online optimization form that mimics the structure of the aforementioned Bellman-optimal policy, where "greediness" is regularized by a measure of residual uncertainty. This new lens of online optimization allows both a better understanding of Thompson Sampling dynamics, as well as a principled manner for policy improvement that mimics the Bellman-optimal benchmark.

---

### EigenScore: OOD Detection using Covariance in Diffusion Models

Authors: Shirin Shoushtari, Yi Wang, Xiao Shi, M. Salman Asif, Ulugbek S. Kamilov
Published: 2025-10-08
Categories: cs.CV
arXiv: https://arxiv.org/abs/2510.07206v1

Out-of-distribution (OOD) detection is critical for the safe deployment of machine learning systems in safety-sensitive domains. Diffusion models have recently emerged as powerful generative models, capable of capturing complex data distributions through iterative denoising. Building on this progress, recent work has explored their potential for OOD detection. We propose EigenScore, a new OOD detection method that leverages the eigenvalue spectrum of the posterior covariance induced by a diffusion model. We argue that posterior covariance provides a consistent signal of distribution shift, leading to larger trace and leading eigenvalues on OOD inputs, yielding a clear spectral signature. We further provide analysis explicitly linking posterior covariance to distribution mismatch, establishing it as a reliable signal for OOD detection. To ensure tractability, we adopt a Jacobian-free subspace iteration method to estimate the leading eigenvalues using only forward evaluations of the denoiser. Empirically, EigenScore achieves SOTA performance, with up to 5% AUROC improvement over the best baseline. Notably, it remains robust in near-OOD settings such as CIFAR-10 vs CIFAR-100, where existing diffusion-based methods often fail.

---

### Guided by the Experts: Provable Feature Learning Dynamic of Soft-Routed Mixture-of-Experts

Authors: Fangshuo Liao, Anastasios Kyrillidis
Published: 2025-10-08
Categories: cs.LG, math.OC
arXiv: https://arxiv.org/abs/2510.07205v1

Mixture-of-Experts (MoE) architectures have emerged as a cornerstone of modern AI systems. In particular, MoEs route inputs dynamically to specialized experts whose outputs are aggregated through weighted summation. Despite their widespread application, theoretical understanding of MoE training dynamics remains limited to either separate expert-router optimization or only top-1 routing scenarios with carefully constructed datasets. This paper advances MoE theory by providing convergence guarantees for joint training of soft-routed MoE models with non-linear routers and experts in a student-teacher framework. We prove that, with moderate over-parameterization, the student network undergoes a feature learning phase, where the router's learning process is ``guided'' by the experts, that recovers the teacher's parameters. Moreover, we show that a post-training pruning can effectively eliminate redundant neurons, followed by a provably convergent fine-tuning process that reaches global optimality. To our knowledge, our analysis is the first to bring novel insights in understanding the optimization landscape of the MoE architecture.

---

### Sunflower: A New Approach To Expanding Coverage of African Languages in Large Language Models

Authors: Benjamin Akera, Evelyn Nafula Ouma, Gilbert Yiga, Patrick Walukagga, Phionah Natukunda, Trevor Saaka, Solomon Nsumba, Lilian Teddy Nabukeera, Joel Muhanguzi, Imran Sekalala, Nimpamya Janat Namara, Engineer Bainomugisha et al.
Published: 2025-10-08
Categories: cs.CL
arXiv: https://arxiv.org/abs/2510.07203v1

There are more than 2000 living languages in Africa, most of which have been bypassed by advances in language technology. Current leading LLMs exhibit strong performance on a number of the most common languages (e.g. Swahili or Yoruba), but prioritise support for the languages with the most speakers first, resulting in piecemeal ability across disparate languages. We contend that a regionally focussed approach is more efficient, and present a case study for Uganda, a country with high linguistic diversity. We describe the development of Sunflower 14B and 32B, a pair of models based on Qwen 3 with state of the art comprehension in the majority of all Ugandan languages. These models are open source and can be used to reduce language barriers in a number of important practical applications.

---

### An in-depth look at approximation via deep and narrow neural networks

Authors: Joris Dommel, Sven A. Wegner
Published: 2025-10-08
Categories: cs.LG
arXiv: https://arxiv.org/abs/2510.07202v1

In 2017, Hanin and Sellke showed that the class of arbitrarily deep, real-valued, feed-forward and ReLU-activated networks of width w forms a dense subset of the space of continuous functions on R^n, with respect to the topology of uniform convergence on compact sets, if and only if w>n holds. To show the necessity, a concrete counterexample function f:R^n->R was used. In this note we actually approximate this very f by neural networks in the two cases w=n and w=n+1 around the aforementioned threshold. We study how the approximation quality behaves if we vary the depth and what effect (spoiler alert: dying neurons) cause that behavior.

---

### Accelerating Inference for Multilayer Neural Networks with Quantum Computers

Authors: Arthur G. Rattew, Po-Wei Huang, Naixu Guo, Lirandë Pira, Patrick Rebentrost
Published: 2025-10-08
Categories: quant-ph, cs.LG
arXiv: https://arxiv.org/abs/2510.07195v2

Fault-tolerant Quantum Processing Units (QPUs) promise to deliver exponential speed-ups in select computational tasks, yet their integration into modern deep learning pipelines remains unclear. In this work, we take a step towards bridging this gap by presenting the first fully-coherent quantum implementation of a multilayer neural network with non-linear activation functions. Our constructions mirror widely used deep learning architectures based on ResNet, and consist of residual blocks with multi-filter 2D convolutions, sigmoid activations, skip-connections, and layer normalizations. We analyse the complexity of inference for networks under three quantum data access regimes. Without any assumptions, we establish a quadratic speedup over classical methods for shallow bilinear-style networks. With efficient quantum access to the weights, we obtain a quartic speedup over classical methods. With efficient quantum access to both the inputs and the network weights, we prove that a network with an $N$-dimensional vectorized input, $k$ residual block layers, and a final residual-linear-pooling layer can be implemented with an error of $ε$ with $O(\text{polylog}(N/ε)^k)$ inference cost.

---

### Covert Quantum Learning: Privately and Verifiably Learning from Quantum Data

Authors: Abhishek Anand, Matthias C. Caro, Ari Karchmer, Saachi Mutreja
Published: 2025-10-08
Categories: quant-ph, cs.CR, cs.LG
arXiv: https://arxiv.org/abs/2510.07193v1

Quantum learning from remotely accessed quantum compute and data must address two key challenges: verifying the correctness of data and ensuring the privacy of the learner's data-collection strategies and resulting conclusions. The covert (verifiable) learning model of Canetti and Karchmer (TCC 2021) provides a framework for endowing classical learning algorithms with such guarantees. In this work, we propose models of covert verifiable learning in quantum learning theory and realize them without computational hardness assumptions for remote data access scenarios motivated by established quantum data advantages. We consider two privacy notions: (i) strategy-covertness, where the eavesdropper does not gain information about the learner's strategy; and (ii) target-covertness, where the eavesdropper does not gain information about the unknown object being learned. We show: Strategy-covert algorithms for making quantum statistical queries via classical shadows; Target-covert algorithms for learning quadratic functions from public quantum examples and private quantum statistical queries, for Pauli shadow tomography and stabilizer state learning from public multi-copy and private single-copy quantum measurements, and for solving Forrelation and Simon's problem from public quantum queries and private classical queries, where the adversary is a unidirectional or i.i.d. ancilla-free eavesdropper. The lattermost results in particular establish that the exponential separation between classical and quantum queries for Forrelation and Simon's problem survives under covertness constraints. Along the way, we design covert verifiable protocols for quantum data acquisition from public quantum queries which may be of independent interest. Overall, our models and corresponding algorithms demonstrate that quantum advantages are privately and verifiably achievable even with untrusted, remote data.

---

### Poisoning Attacks on LLMs Require a Near-constant Number of Poison Samples

Authors: Alexandra Souly, Javier Rando, Ed Chapman, Xander Davies, Burak Hasircioglu, Ezzeldin Shereen, Carlos Mougan, Vasilios Mavroudis, Erik Jones, Chris Hicks, Nicholas Carlini, Yarin Gal et al.
Published: 2025-10-08
Categories: cs.LG
arXiv: https://arxiv.org/abs/2510.07192v1

Poisoning attacks can compromise the safety of large language models (LLMs) by injecting malicious documents into their training data. Existing work has studied pretraining poisoning assuming adversaries control a percentage of the training corpus. However, for large models, even small percentages translate to impractically large amounts of data. This work demonstrates for the first time that poisoning attacks instead require a near-constant number of documents regardless of dataset size. We conduct the largest pretraining poisoning experiments to date, pretraining models from 600M to 13B parameters on chinchilla-optimal datasets (6B to 260B tokens). We find that 250 poisoned documents similarly compromise models across all model and dataset sizes, despite the largest models training on more than 20 times more clean data. We also run smaller-scale experiments to ablate factors that could influence attack success, including broader ratios of poisoned to clean data and non-random distributions of poisoned samples. Finally, we demonstrate the same dynamics for poisoning during fine-tuning. Altogether, our results suggest that injecting backdoors through data poisoning may be easier for large models than previously believed as the number of poisons required does not scale up with model size, highlighting the need for more research on defences to mitigate this risk in future models.

---

### Resolution scaling governs DINOv3 transfer performance in chest radiograph classification

Authors: Soroosh Tayebi Arasteh, Mina Shaigan, Christiane Kuhl, Jakob Nikolas Kather, Sven Nebelung, Daniel Truhn
Published: 2025-10-08
Categories: cs.CV, cs.AI, cs.LG
arXiv: https://arxiv.org/abs/2510.07191v3

Self-supervised learning (SSL) has improved visual representation learning, but its value in chest radiography remains uncertain. DINOv3 extends earlier SSL models through Gram-anchored self-distillation and explicit high-resolution adaptation. Whether these changes improve transfer learning for chest radiograph classification has not been established. We benchmarked DINOv3 against DINOv2 and supervised ImageNet initialization across seven chest radiograph datasets comprising 816,183 radiographs from pediatric and adult cohorts. ViT-B/16 and ConvNeXt-B were evaluated under full fine-tuning at 224 and 512 pixels, with targeted 1024 experiments on three cohorts. Additional analyses examined parameter-efficient adaptation, synthetic label corruption, external validation, frozen 7B features, and computational efficiency. The primary outcome was mean AUROC across labels. In adult cohorts, DINOv3 did not consistently outperform DINOv2 at 224 x 224 pixels, but became the strongest initialization at 512 x 512, especially with ConvNeXt-B. Gains were greatest for small focal and boundary-dependent abnormalities, whereas large-structure findings changed little. The pediatric cohort showed no significant benefit from DINOv3, higher resolution, or backbone choice. Scaling to 1024 x 1024 rarely improved performance and markedly increased computational cost. ConvNeXt-B remained superior to ViT-B/16 under both full and parameter-efficient adaptation. External validation preserved the 512 x 512 DINOv3 advantage, whereas synthetic label corruption showed that this benefit should not be interpreted simply as superior noise robustness. For adult chest radiograph classification, DINOv3 provides its most reliable benefit at 512 x 512 pixels, particularly with ConvNeXt-B. Fully adapted mid-sized models at 512 x 512 pixels provided the best performance-cost trade-off in our benchmark.

---

### MV-Performer: Taming Video Diffusion Model for Faithful and Synchronized Multi-view Performer Synthesis

Authors: Yihao Zhi, Chenghong Li, Hongjie Liao, Xihe Yang, Zhengwentai Sun, Jiahao Chang, Xiaodong Cun, Wensen Feng, Xiaoguang Han
Published: 2025-10-08
Categories: cs.CV
arXiv: https://arxiv.org/abs/2510.07190v1

Recent breakthroughs in video generation, powered by large-scale datasets and diffusion techniques, have shown that video diffusion models can function as implicit 4D novel view synthesizers. Nevertheless, current methods primarily concentrate on redirecting camera trajectory within the front view while struggling to generate 360-degree viewpoint changes. In this paper, we focus on human-centric subdomain and present MV-Performer, an innovative framework for creating synchronized novel view videos from monocular full-body captures. To achieve a 360-degree synthesis, we extensively leverage the MVHumanNet dataset and incorporate an informative condition signal. Specifically, we use the camera-dependent normal maps rendered from oriented partial point clouds, which effectively alleviate the ambiguity between seen and unseen observations. To maintain synchronization in the generated videos, we propose a multi-view human-centric video diffusion model that fuses information from the reference video, partial rendering, and different viewpoints. Additionally, we provide a robust inference procedure for in-the-wild video cases, which greatly mitigates the artifacts induced by imperfect monocular depth estimation. Extensive experiments on three datasets demonstrate our MV-Performer's state-of-the-art effectiveness and robustness, setting a strong model for human-centric 4D novel view synthesis.

---

### Split Conformal Classification with Unsupervised Calibration

Authors: Santiago Mazuelas
Published: 2025-10-08
Categories: stat.ML, cs.LG
arXiv: https://arxiv.org/abs/2510.07185v2

Methods for split conformal prediction leverage calibration samples to transform any prediction rule into a set-prediction rule that complies with a target coverage probability. Existing methods provide remarkably strong performance guarantees with minimal computational costs. However, they require to use calibration samples composed by labeled examples different to those used for training. This requirement can be highly inconvenient, as it prevents the use of all labeled examples for training and may require acquiring additional labels solely for calibration. This paper presents an effective methodology for split conformal prediction with unsupervised calibration for classification tasks. In the proposed approach, set-prediction rules are obtained using unsupervised calibration samples together with supervised training samples previously used to learn the classification rule. Theoretical and experimental results show that the presented methods can achieve performance comparable to that with supervised calibration, at the expenses of a moderate degradation in performance guarantees and computational efficiency.

---

### Bridged Clustering: Semi-Supervised Sparse Bridging

Authors: Patrick Peixuan Ye, Chen Shani, Ellen Vitercik
Published: 2025-10-08
Categories: cs.LG
arXiv: https://arxiv.org/abs/2510.07182v3

We introduce Bridged Clustering, a semi-supervised framework to learn predictors from any unpaired input $X$ and output $Y$ dataset. Our method first clusters $X$ and $Y$ independently, then learns a sparse, interpretable bridge between clusters using only a few paired examples. At inference, a new input $x$ is assigned to its nearest input cluster, and the centroid of the linked output cluster is returned as the prediction $\hat{y}$. Unlike traditional SSL, Bridged Clustering explicitly leverages output-only data, and unlike dense transport-based methods, it maintains a sparse and interpretable alignment. Through theoretical analysis, we show that with bounded mis-clustering and mis-bridging rates, our algorithm becomes an effective and efficient predictor. Empirically, our method is competitive with SOTA methods while remaining simple, model-agnostic, and highly label-efficient in low-supervision settings.

---

### TIGeR: Tool-Integrated Geometric Reasoning in Vision-Language Models for Robotics

Authors: Yi Han, Enshen Zhou, Shanyu Rong, Jingkun An, Pengwei Wang, Zhongyuan Wang, Cheng Chi, Lu Sheng, Shanghang Zhang
Published: 2025-10-08
Categories: cs.RO, cs.AI, cs.CV
arXiv: https://arxiv.org/abs/2510.07181v3

Vision-Language Models (VLMs) have shown remarkable capabilities in spatial reasoning, yet they remain fundamentally limited to qualitative precision and lack the computational precision required for real-world robotics. Current approaches fail to leverage metric cues from depth sensors and camera calibration, instead reducing geometric problems to pattern recognition tasks that cannot deliver the centimeter-level accuracy essential for robotic manipulation. We present TIGeR (Tool-Integrated Geometric Reasoning), a novel framework that transforms VLMs from perceptual estimators to geometric computers by enabling them to generate and execute precise geometric computations through external tools. Rather than attempting to internalize complex geometric operations within neural networks, TIGeR empowers models to recognize geometric reasoning requirements, synthesize appropriate computational code, and invoke specialized libraries for exact calculations. To support this paradigm, we introduce TIGeR-300K, a comprehensive tool-invocation-oriented dataset covering point transformations, pose estimation, and spatial compatibility verification, complete with tool invocation sequences and intermediate computations. Through a two-stage training pipeline combining supervised fine-tuning (SFT) and reinforcement fine-tuning (RFT) with our proposed hierarchical reward design, TIGeR achieves SOTA performance on geometric reasoning benchmarks while demonstrating centimeter-level precision in real-world robotic manipulation tasks.

---

### Bayesian Portfolio Optimization by Predictive Synthesis

Authors: Masahiro Kato, Kentaro Baba, Hibiki Kaibuchi, Ryo Inokuchi
Published: 2025-10-08
Categories: econ.EM, cs.LG, q-fin.CP, q-fin.PM, stat.AP
arXiv: https://arxiv.org/abs/2510.07180v1

Portfolio optimization is a critical task in investment. Most existing portfolio optimization methods require information on the distribution of returns of the assets that make up the portfolio. However, such distribution information is usually unknown to investors. Various methods have been proposed to estimate distribution information, but their accuracy greatly depends on the uncertainty of the financial markets. Due to this uncertainty, a model that could well predict the distribution information at one point in time may perform less accurately compared to another model at a different time. To solve this problem, we investigate a method for portfolio optimization based on Bayesian predictive synthesis (BPS), one of the Bayesian ensemble methods for meta-learning. We assume that investors have access to multiple asset return prediction models. By using BPS with dynamic linear models to combine these predictions, we can obtain a Bayesian predictive posterior about the mean rewards of assets that accommodate the uncertainty of the financial markets. In this study, we examine how to construct mean-variance portfolios and quantile-based portfolios based on the predicted distribution information.

---

### Biasless Language Models Learn Unnaturally: How LLMs Fail to Distinguish the Possible from the Impossible

Authors: Imry Ziv, Nur Lan, Emmanuel Chemla
Published: 2025-10-08
Categories: cs.CL
arXiv: https://arxiv.org/abs/2510.07178v2

Are large language models (LLMs) sensitive to the distinction between humanly possible and impossible languages? This question was recently used in a broader debate on whether LLMs and humans share the same innate learning biases. Previous work has answered it in the positive by comparing LLM learning curves on existing language datasets and on "impossible" datasets derived from them via various perturbation functions. Using the same methodology, we examine this claim on a wider set of languages and impossible perturbations. We find that in most cases, GPT-2 learns each language and its impossible counterpart equally easily, in contrast to previous findings. We also apply a more lenient condition by testing whether GPT-2 provides any kind of separation between the whole sets of natural vs. impossible languages, based on cross-linguistic variance in metrics derived from the learning curves. Taken together, these perspectives show that GPT-2 provides no systematic separation between the possible and the impossible.

---

### CARPAS: Towards Content-Aware Refinement of Provided Aspects for Summarization in Large Language Models

Authors: Yong-En Tian, Yu-Chien Tang, An-Zi Yen, Wen-Chih Peng
Published: 2025-10-08
Categories: cs.CL
arXiv: https://arxiv.org/abs/2510.07177v2

Aspect-based summarization has attracted significant attention for its ability to generate more fine-grained and user-aligned summaries. While most existing approaches assume a set of predefined aspects as input, real-world scenarios often present challenges where these given aspects may be incomplete, irrelevant, or entirely missing from the document. Users frequently expect systems to adaptively refine or filter the provided aspects based on the actual content. In this paper, we initiate this novel task setting, termed Content-Aware Refinement of Provided Aspects for Summarization (CARPAS), with the aim of dynamically adjusting the provided aspects based on the document context before summarizing. We construct three new datasets and conduct pilot experiments with four representative prompting strategies. Our analysis reveals that LLMs often over-generate aspects, resulting in excessively long and misaligned summaries. Building on this observation, we introduce a two-stage framework that derives lightweight scope guidance before aspect refinement and summarization, improving focus and reducing over-generation. Our extensive experiments show that the proposed approach significantly improves performance across all datasets. Moreover, our deeper analyses uncover LLMs' compliance when the requested number of aspects differs from their own estimations, establishing a crucial insight for the deployment of LLMs in similar real-world applications.

---