The paper investigates whether public medical vision‑language benchmarks contain data that may have leaked into the training sets of open VLMs. Using four contamination detectors, the authors find that a significant portion of images in SLAKE‑En have close neighbours in PMC‑OA‑beta, indicating source overlap rather than exact memorization. On the text side, only the exchangeability detector consistently flags potential contamination, while the other methods produce false positives and are unreliable without external baselines.
By Bruce Changlong Xu, Lan Wu, Alexander Ryu
arXiv:2610. 30518v1 Announce Type: new Abstract: We present Protegrity-PII, a new PII decision model for structured and unstructured text.
By Hardeep Bassi, Hui Ma, Lu\'is Seabra, Koowum Joshi, Daniel Pereira, Averell Gatton
ULME is a post‑training framework that unifies text embedding and language modeling within a single causal decoder‑only LLM by inserting learnable slots for semantic and reasoning embeddings. It trains three objectives—contrastive loss for semantic alignment, language‑modeling loss for answer generation, and a stop‑gradient predictive loss linking reasoning to the answer—so that generation actively shapes the embedding. Across Qwen2.5 and Qwen3 models (0.5B–8B), ULME improves MTEB scores by 3.3–4.1 points, tightens BRIGHT and AdvBench‑IR, and matches generation‑only fine‑tunes on standard accuracy benchmarks.
By Yang Liu, Zhiqiang Yuan, Shu-Dong Huang, Jiancheng Lv, Liwei Yang, Qiwei Yan, Kailin Lv, Jinchao Zhang
The paper investigates natural language autoencoders (NLAs) as tools for verbalizing activation differences in large language models. It examines whether NLAs capture compositional structure, generalize to fine‑tuned models, recover fine‑tuning objectives, and can be specialized for model diffing. Experiments across four models show that NLAs largely succeed in these tasks, though they do not yet match state‑of‑the‑art diffing methods on emergent misalignment detection.
By Andrea Blasi N\'u\~nez, Gianluca Barmina, Federico Torrielli, Thiusius Rajeeth Savarimuthu, Peter Schneider-Kamp, Lukas Galke Poech
The paper reexamines GAN training by replacing the traditional cross‑entropy discriminator with an upper bound on the Bayes error rate, using the Bayes optimal learning threshold (BOLT) loss. It shows that, with balanced class priors, the resulting adversarial game is equivalent to minimizing total variation between data and generator distributions, and introduces BOLT‑GAN when the discriminator is constrained to be 1‑Lipschitz. Experiments on four image‑generation benchmarks demonstrate that BOLT‑GAN achieves better sample quality and coverage than WGAN‑GP, reducing FID scores on CIFAR‑10 and LSUN Bedroom‑64 within a 20‑epoch budget.
By Mohammadreza Tavasoli Naeini, Ali Bereyhi, Morteza Noshad, Ben Liang, Alfred O. Hero III
The paper proposes treating few-shot classification as a series of multiple instance verification (MIV) tasks, enabling effective transfer across domains without fine-tuning the backbone. It introduces the MIV-head, a backbone‑agnostic, computationally efficient classification head that achieves competitive accuracy on cross‑domain few‑shot image classification using frozen, pre‑trained backbones. Experiments on an extended Meta‑Dataset benchmark show that the MIV-head outperforms state‑of‑the‑art adapter methods while incurring lower adaptation cost.
By Xin Xu, Eibe Frank, Geoffrey Holmes
AtManRL introduces differentiable attention manipulation to train salient chain-of-thought reasoning in large language models. By optimizing an additive attention mask that identifies the tokens necessary to recover the model’s answer, the method derives a saliency reward combined with an outcome reward in a GRPO framework with decoupled advantages. Across three models and four benchmarks, AtManRL improves saliency and faithfulness metrics while maintaining comparable accuracy to length‑penalty baselines.
By Max Henning H\"oth, Kristian Kersting, Bj\"orn Deiseroth, Letitia Parcalabescu
The paper introduces a minimal‑edit on‑policy distillation method for vision‑language models to correct misread diagrams in visual mathematical reasoning. By using reference figure descriptions and verified solutions only during training, the teacher evaluates student rollouts and penalizes only the tokens that deviate from correct reasoning, while correct rollouts receive reinforcement learning rewards. The approach eliminates the need for figure descriptions at test time, improves low‑level geometric perception, and transfers well to external math and perception benchmarks.
By Wen Ye, Ailin Deng, Zirui Wang, Shixing Yu, Han Zhang, Tian Yun, Nikhil Mehta, Tarun Kalluri, Xuefei Cao, Xuezhe Ma, Hengduo Li
The paper introduces Domain Prompt Learning (DoP), a zero‑shot domain generalization method that leverages a large vision‑language model like CLIP to recognize unseen classes in unseen target domains without any source‑domain data. DoP learns domain knowledge through text‑domain learning (TDL), which generates diverse style tokens in the text space, and visual‑domain learning (VDL), which refines these tokens using test visual data via self‑supervised learning. Experiments on DG benchmarks such as PACS, VLCS, OfficeHome, DomainNet, and ImageNet demonstrate that DoP consistently outperforms state‑of‑the‑art DG approaches.
By Shiming Chen, Shuhuang Chen, Salman Khan, Fahad Shahbaz Khan
PROACT-Agent is a framework designed to provide proactive runtime oversight for large language model agents, aiming to prevent irreversible environmental harm by enabling real‑time guardrails. It introduces three key components: Progressive Trajectory Unrolling, Reasoning‑Augmented Causal Rectification, and Culturally‑Aware Data Localization, which together address safety drift and enforce temporal consistency. The accompanying PROACT‑Bench, a bilingual safety benchmark with 155,780 labeled states, demonstrates the framework’s effectiveness, achieving high unsafe‑class F1 scores and significantly reducing attack success rates in AgentDojo.
By Ding Jia, Wei Liu, Xianglong Du, Yingjie Li, Yingqing Yang, Huili Yu, Zhangsong Zhan, Chu Zhou
The paper surveys the growing integration of Generative Artificial Intelligence (GenAI) with the Internet of Things (IoT), noting its potential to reshape sectors such as Healthcare, Manufacturing, and Smart Cities. It highlights that most current studies focus on narrow applications, underscoring the need for a broader analysis of GenAI‑IoT synergy. The authors conduct a systematic literature review using PRISMA, present a comparison framework, and outline research questions to guide future exploration of GenAI in IoT computing.
By Fabrizio Mangione, Claudio Savaglio, Giancarlo Fortino
The paper introduces FedGRC, a federated learning framework that enhances both global recognition and personalization for heterogeneous clients. It combines Consensus‑Guided Representation Learning, Multilayer Metric Synthesis, and Support‑Adaptive Shrinkage Personalization to address failures on rare or unseen classes. Experiments on CIFAR‑10, CIFAR‑100, and the FedRS dataset show that FedGRC outperforms 21 baseline methods, improving global accuracy and personalization floors by over 2% in several settings.
By Junyu Lin, Bosong Qi, Pengcheng Shi, Jiayuan Li
UserSWE is a new benchmark that evaluates coding agents’ ability to fix repository-level bugs through multi‑turn conversations with dynamic users, providing execution‑grounded verdicts after each turn. The benchmark transforms repository histories and existing single‑turn datasets into multi‑turn tasks, using a persona chain derived from 20,859 development sessions to simulate user interactions. Eleven agents were tested on 72 tasks from 42 repositories, revealing that agent rankings shift with the interaction budget and that single‑turn performance alone does not capture interactive coding capability.
By Zhan-Yu Wu, Ji Luo, Alyssa Wu, Junjielong Xu, Richeng Xuan, Zhichao Hu, Chongming Gao, Xinting Hu
arXiv:2610. 30072v1 Announce Type: new Abstract: Data agents built on ReAct-style loops now execute complex analytical workflows over structured data, yet their intermediate reasoning still runs almost entirely through text.
By Xinzhe Wang, Xudong Weng, Zang Li, Shaolei Zhang
The paper introduces VHDLBench, a benchmark and fine‑tuning framework for large language models (LLMs) applied to VHDL-based chip design. It presents a dataset of over 18,500 syntax‑verified VHDL examples, fine‑tunes five 7B‑parameter LLMs with QLoRA, and evaluates them through a three‑level pipeline—syntax/synthesis validation, functional simulation, and FPGA prototyping on Zynq SoCs. Results show up to 92% syntax correctness and 73% functional accuracy, with DeepSeek‑Coder achieving the best performance and successful deployment on real hardware.
By Patrik Drazic, Benaoumeur Senouci, Boualem Benatallah
The paper introduces a method for certifying the quality of solutions produced by heuristic solvers for general‑sum extensive‑form games. By relating the Nash gap of a weighted average strategy to external regret and a signed correction term, the authors derive an online certificate that can be computed in linear time per checkpoint. Experiments on the biased Leduc family show that selecting the iterate with the smallest certificate from a portfolio of training‑free regret‑matching solvers achieves high success rates across multiple test sets.
By Boning Li, Longbo Huang
The paper introduces an open generator for telecom alarm storms that produces exact root causes and propagation paths for every alarm on real backbone topologies, with each emitted field tied to standardized terms from NORIA‑O, RFC 8632, RFC 8345, or TMF642. Unlike previous benchmarks, the generator does not rely on confidential operator logs and includes fixed configurations, splits, a scorer, and nine baselines. A simple rule—ranking the device that raised the earliest critical alarm—outperforms more complex methods on both synthetic and real operator data, highlighting a key property of alarm streams.
By Madhulatha Mandarapu, Sandeep Kunkunuru
The paper reports how an AI-driven system, AI-greement, was used to discover a new consensus protocol called A‑agreement for geo‑replicated state machines. A‑agreement improves mean latency over existing baselines by 12.2% on write‑heavy and 33.3% on read‑heavy workloads while matching performance on read‑only workloads, using lease‑authorized local reads, direct client learning for leader‑ordered writes, and a fast fixed‑quorum per‑key write path. The AI‑agreement harness integrates a protocol knowledge base, a modular implementation skeleton, and a Lean verification framework, revealing vulnerabilities in existing protocols and demonstrating how domain‑specific harnesses can transform general‑purpose coding agents into effective tools for distributed computing research.
By Cl\'ement Burgelin, Rachid Guerraoui, Zicong Hong, Yinan Hu, Anne-Marie Kermarrec, Rafael Pires, Rasmus Moorits Veski, Martijn de Vos
The paper introduces a quantum‑inspired data imputation framework that combines Principal Component Analysis with quantum rotation‑based reconstruction. Rotation angles are chosen using gradient‑free optimizers (Simulated Annealing, COBYLA, Differential Evolution) and the top three solutions per method are retained to reconstruct missing values within realistic bounds. Benchmarks on the UCI Pima Indians Diabetes dataset show that the quantum‑inspired variants achieve substantially lower Wasserstein W1 distances and better preserve variance compared to classical and modern baseline imputers.
By Nishikanta Mohanty, Bikash K. Behera, Badshah Mukherjee, Pravat Dash, Christopher Ferrie
The paper investigates whether matching the spectral predictability Ω of synthetic pretraining data to a target dataset is sufficient for effective transfer learning under distribution shift. Experiments across LSTM, PatchTST, and TCN architectures show that Ω-matched pretraining can outperform an oracle baseline in synthetic-to-synthetic settings, but on the real-world ETTm1 benchmark it offers only marginal gains or even degrades performance, with Ω providing little guidance on beneficial source selection. The authors conclude that while Ω matching can help in controlled scenarios, it is not a reliable predictor of transfer success and other signal properties must be considered.
By Edric Castel C. Hao, Juan Raphael S. Se\~na, Neon S. De Jesus