Fresh from the feed

arXiv:2510.12732v2 Announce Type: replace Abstract: JavaScript engines are widely used in web browsers, PDF readers, and server-side applications. The rise in concern over their security has led to the development of several targeted fuzzing techniques. However, existing approaches use random selection to determine where to perform mutations in JavaScript code. We postulate that the problem of selecting better mutation targets is suitable for combinatorial bandits with a volatile number of arms. Thus, we propose CLUTCH, a novel deep combinatorial bandit that can observe variable length JavaScript test case representations, using an attention mechanism from deep learning. Furthermore, using Concrete Dropout, CLUTCH can dynamically adapt its exploration. We show that CLUTCH increases efficiency in JavaScript fuzzing compared to three state-of-the-art solutions by increasing the number of valid test cases and coverage-per-testcase by, respectively, 20.3% and 8.9% on average. In volatile and combinatorial settings we show that CLUTCH outperforms state-of-the-art bandits, achieving at least 78.1% and 4.1% less regret in volatile and combinatorial settings, respectively.

arXiv:2508.03404v2 Announce Type: replace-cross Abstract: The dominant paradigm of monolithic scaling in Vision-Language Models (VLMs) is failing for understanding and reasoning in documents, yielding diminishing returns as it struggles with the inherent need of this domain for document-based procedural reasoning, cognitive complexity, and factual accuracy. To this end, we introduce MACT, a Multi-Agent Collaboration framework with agent-wise adaptive Test-time scaling that pioneers a paradigm shift to procedural scaling, adapting dynamically to the functional entities of visual documents understanding and reasoning. MACT decomposes the visual document processing flow into four specialized agents, i.e., planning, execution, judgment, and answer, to resolve cognitive overload and introduce a critical self-correction loop for factual grounding. This collaborative architecture is amplified by an agent-wise adaptive test-time scaling strategy that intelligently allocates computational resources based on the complexity and redundancy of each functionality. Evaluated on multiple visual document understanding benchmarks, MACT achieves superior performance with a smaller parameter scale, adapting effectively to various document scenarios without compromising its general or mathematical reasoning capabilities. The three variants of MACT consistently attain top-three average performance rankings, with average performance enhancements of 9.9-11.5% over the base models. The source code will be released publicly.

arXiv:2511.11030v2 Announce Type: cross Abstract: Artificial intelligence is revealing what medicine never intended to encode. Deep vision models, trained on chest X-rays, can now detect not only disease but also invisible traces of social inequality. In this study, we show that state-of-the-art architectures (DenseNet121, SwinV2-B, MedMamba) can predict a patient's health insurance type, a strong proxy for socioeconomic status, from normal chest X-rays with significant accuracy (AUC around 0.67 on MIMIC-CXR-JPG, 0.68 on CheXpert). The signal persists even when age, race, and sex are controlled for, and remains detectable when the model is trained exclusively on a single racial group. Patch-based occlusion reveals that the signal is diffuse rather than localized, embedded in the upper and mid-thoracic regions. This suggests that deep networks may be internalizing subtle traces of clinical environments, equipment differences, or care pathways; learning socioeconomic segregation itself. These findings challenge the assumption that medical images are neutral biological data. By uncovering how models perceive and exploit these hidden social signatures, this work reframes fairness in medical AI: the goal is no longer only to balance datasets or adjust thresholds, but to interrogate and disentangle the social fingerprints embedded in clinical data itself.

arXiv:2511.11041v1 Announce Type: cross Abstract: We find that current text embedding models produce outputs with a consistent bias, i.e., each embedding vector $e$ can be decomposed as $\tilde{e} + \mu$, where $\mu$ is almost identical across all sentences. We propose a plug-and-play, training-free and lightweight solution called Renormalization. Through extensive experiments, we show that renormalization consistently and statistically significantly improves the performance of existing models on the Massive Multilingual Text Embedding Benchmark (MMTEB). In particular, across 38 models, renormalization improves performance by 9.7 $\sigma$ on retrieval tasks, 3.1 $\sigma$ on classification tasks, and 0.8 $\sigma$ on other types of tasks. Renormalization has two variants: directly subtracting $\mu$ from $e$, or subtracting the projection of $e$ onto $\mu$. We theoretically predict that the latter performs better, and our experiments confirm this prediction.

arXiv:2511.11046v1 Announce Type: cross Abstract: Graph neural networks (GNNs) have become an indispensable tool for analyzing relational data. In the literature, classical GNNs may be classified into three variants: convolutional, attentional, and message-passing. While the standard message-passing variant is highly expressive, its typical pair-wise messages nevertheless only consider the features of the center node and each neighboring node individually. This design fails to incorporate the rich contextual information contained within the broader local neighborhood, potentially hindering its ability to learn complex relationships within the entire set of neighboring nodes. To address this limitation, this work first formalizes the concept of neighborhood-contextualization, rooted in a key property of the attentional variant. This then serves as the foundation for generalizing the message-passing variant to the proposed neighborhood-contextualized message-passing (NCMP) framework. To demonstrate its utility, a simple, practical, and efficient method to parametrize and operationalize NCMP is presented, leading to the development of the proposed Soft-Isomorphic Neighborhood-Contextualized Graph Convolution Network (SINC-GCN). A preliminary analysis on a synthetic binary node classification problem then underscores both the expressivity and efficiency of the proposed GNN architecture. Overall, the paper lays the foundation for the novel NCMP framework as a practical path toward further enhancing the graph representational power of classical GNNs.

arXiv:2511.11062v1 Announce Type: cross Abstract: Diffusion Transformers, particularly for video generation, achieve remarkable quality but suffer from quadratic attention complexity, leading to prohibitive latency. Existing acceleration methods face a fundamental trade-off: dynamically estimating sparse attention patterns at each denoising step incurs high computational overhead and estimation errors, while static sparsity patterns remain fixed and often suboptimal throughout denoising. We identify a key structural property of diffusion attention, namely, its sparsity patterns exhibit strong temporal coherence across denoising steps. Tiles deemed non-essential at step $t$ typically remain so at step $t+\delta$. Leveraging this observation, we introduce LiteAttention, a method that exploits temporal coherence to enable evolutionary computation skips across the denoising sequence. By marking non-essential tiles early and propagating skip decisions forward, LiteAttention eliminates redundant attention computations without repeated profiling overheads, combining the adaptivity of dynamic methods with the efficiency of static ones. We implement a highly optimized LiteAttention kernel on top of FlashAttention and demonstrate substantial speedups on production video diffusion models, with no degradation in quality. The code and implementation details will be publicly released.

arXiv:2510.09616v2 Announce Type: replace-cross Abstract: Industrial Control Systems (ICS) in water distribution and treatment face cyber-physical attacks exploiting network and physical vulnerabilities. Current water system anomaly detection methods rely on correlations, yielding high false alarms and poor root cause analysis. We propose a Causal Digital Twin (CDT) framework for water infrastructures, combining causal inference with digital twin modeling. CDT supports association for pattern detection, intervention for system response, and counterfactual analysis for water attack prevention. Evaluated on water-related datasets SWaT, WADI, and HAI, CDT shows 90.8\% compliance with physical constraints and structural Hamming distance 0.133 $\pm$ 0.02. F1-scores are $0.944 \pm 0.014$ (SWaT), $0.902 \pm 0.021$ (WADI), $0.923 \pm 0.018$ (HAI, $p<0.0024$). CDT reduces false positives by 74\%, achieves 78.4\% root cause accuracy, and enables counterfactual defenses reducing attack success by 73.2\%. Real-time performance at 3.2 ms latency ensures safe and interpretable operation for medium-scale water systems.

arXiv:2511.11066v1 Announce Type: cross Abstract: Radiology Report Generation (RRG) aims to automatically generate diagnostic reports from radiology images. To achieve this, existing methods have leveraged the powerful cross-modal generation capabilities of Multimodal Large Language Models (MLLMs), primarily focusing on optimizing cross-modal alignment between radiographs and reports through Supervised Fine-Tuning (SFT). However, by only performing instance-level alignment with the image-text pairs, the standard SFT paradigm fails to establish anatomically-grounded alignment, where the templated nature of reports often leads to sub-optimal generation quality. To address this, we propose \textsc{S2D-Align}, a novel SFT paradigm that establishes anatomically-grounded alignment by leveraging auxiliary signals of varying granularities. \textsc{S2D-Align} implements a shallow-to-deep strategy, progressively enriching the alignment process: it begins with the coarse radiograph-report pairing, then introduces reference reports for instance-level guidance, and ultimately utilizes key phrases to ground the generation in specific anatomical details. To bridge the different alignment stages, we introduce a memory-based adapter that empowers feature sharing, thereby integrating coarse and fine-grained guidance. For evaluation, we conduct experiments on the public \textsc{MIMIC-CXR} and \textsc{IU X-Ray} benchmarks, where \textsc{S2D-Align} achieves state-of-the-art performance compared to existing methods. Ablation studies validate the effectiveness of our multi-stage, auxiliary-guided approach, highlighting a promising direction for enhancing grounding capabilities in complex, multi-modal generation tasks.

arXiv:2505.08578v2 Announce Type: replace-cross Abstract: Conformal prediction is a popular method to construct prediction intervals for black-box machine learning models with marginal coverage guarantees. In applications with potentially high-impact events, such as flooding or financial crises, regulators often require very high confidence for such intervals. However, if the desired level of confidence is too large relative to the amount of data used for calibration, then classical conformal methods provide infinitely wide, thus, uninformative prediction intervals. In this paper, we propose a new method to overcome this limitation. We bridge extreme value statistics and conformal prediction to provide reliable and informative prediction intervals with high-confidence coverage, which can be constructed using any black-box extreme quantile regression method. A weighted version of our approach can account for nonstationary data. The advantages of our extreme conformal prediction method are illustrated in a simulation study and in an application to flood risk forecasting.

arXiv:2511.11083v2 Announce Type: cross Abstract: Zero-shot coordination(ZSC) has become a hot topic in reinforcement learning research recently. It focuses on the generalization ability of agents, requiring them to coordinate well with collaborators that are not seen before without any fine-tuning. Population-based training has been proven to provide good zero-shot coordination performance; nevertheless, existing methods are limited by computational resources, mainly focusing on optimizing diversity in small populations while neglecting the potential performance gains from scaling population size. To address this issue, this paper proposes the Scalable Population Training (ScaPT), an efficient training framework comprising two key components: a meta-agent that efficiently realizes a population by selectively sharing parameters across agents, and a mutual information regularizer that guarantees population diversity. To empirically validate the effectiveness of ScaPT, this paper evaluates it along with representational frameworks in Hanabi and confirms its superiority.

arXiv:2511.11299v1 Announce Type: cross Abstract: Multimodal Large Language Models (MLLMs) achieve impressive performance once optimized on massive datasets. Such datasets often contain sensitive or copyrighted content, raising significant data privacy concerns. Regulatory frameworks mandating the 'right to be forgotten' drive the need for machine unlearning. This technique allows for the removal of target data without resource-consuming retraining. However, while well-studied for text, visual concept unlearning in MLLMs remains underexplored. A primary challenge is precisely removing a target visual concept without disrupting model performance on related entities. To address this, we introduce AUVIC, a novel visual concept unlearning framework for MLLMs. AUVIC applies adversarial perturbations to enable precise forgetting. This approach effectively isolates the target concept while avoiding unintended effects on similar entities. To evaluate our method, we construct VCUBench. It is the first benchmark designed to assess visual concept unlearning in group contexts. Experimental results demonstrate that AUVIC achieves state-of-the-art target forgetting rates while incurs minimal performance degradation on non-target concepts.

arXiv:2511.11311v1 Announce Type: cross Abstract: The field of computer vision is undergoing a paradigm shift toward large-scale foundation model pre-training via self-supervised learning (SSL). Leveraging large volumes of unlabeled brain MRI data, such models can learn anatomical priors that improve few-shot performance in diverse neuroimaging tasks. However, most SSL frameworks are tailored to natural images, and their adaptation to capture multi-modal MRI information remains underexplored. This work proposes a modality-invariant representation learning setup and evaluates its effectiveness in stroke and epilepsy lesion segmentation, following large-scale pre-training. Experimental results suggest that despite successful cross-modality alignment, lesion segmentation primarily benefits from preserving fine-grained modality-specific features. Model checkpoints and code are made publicly available.

arXiv:2511.11315v1 Announce Type: cross Abstract: Natural Language Processing (NLP) has transformed the financial industry, enabling advancements in areas such as textual analysis, risk management, and forecasting. Large language models (LLMs) like BloombergGPT and FinMA have set new benchmarks across various financial NLP tasks, including sentiment analysis, stock movement prediction, and credit risk assessment. Furthermore, FinMA-ES, a bilingual financial LLM, has also demonstrated strong performance using the FLARE and FLARE-ES benchmarks. However, the high computational demands of these models limit the accessibility of many organizations. To address this, we propose Layer-wise Adaptive Ensemble Tuning (LAET), a novel strategy that selectively fine-tunes the most effective layers of pre-trained LLMs by analyzing hidden state representations while freezing less critical layers. LAET significantly reduces computational overhead while enhancing task-specific performance. Our approach shows strong results in financial NLP tasks, outperforming existing benchmarks and state-of-the-art LLMs such as GPT-4, even with smaller LLMs ($\sim$3B parameters). This work bridges cutting-edge financial NLP research and real-world deployment with efficient and scalable models for financial applications.

arXiv:2511.08136v2 Announce Type: replace-cross Abstract: In this work, we study the problem of offline safe imitation learning (IL). In many real-world settings, online interactions can be risky, and accurately specifying the reward and the safety cost information at each timestep can be difficult. However, it is often feasible to collect trajectories reflecting undesirable or risky behavior, implicitly conveying the behavior the agent should avoid. We refer to these trajectories as non-preferred trajectories. Unlike standard IL, which aims to mimic demonstrations, our agent must also learn to avoid risky behavior using non-preferred trajectories. In this paper, we propose a novel approach, SafeMIL, to learn a parameterized cost that predicts if the state-action pair is risky via Multiple Instance Learning. The learned cost is then used to avoid non-preferred behaviors, resulting in a policy that prioritizes safety. We empirically demonstrate that our approach can learn a safer policy that satisfies cost constraints without degrading the reward performance, thereby outperforming several baselines.

arXiv:2511.11347v2 Announce Type: cross Abstract: Retrieval-augmented generation (RAG) has rapidly emerged as a transformative approach for integrating large language models into clinical and biomedical workflows. However, privacy risks, such as protected health information (PHI) exposure, remain inconsistently mitigated. This review provides a thorough analysis of the current landscape of RAG applications in healthcare, including (i) sensitive data type across clinical scenarios, (ii) the associated privacy risks, (iii) current and emerging data-privacy protection mechanisms and (iv) future direction for patient data privacy protection. We synthesize 23 articles on RAG applications in healthcare and systematically analyze privacy challenges through a pipeline-structured framework encompassing data storage, transmission, retrieval and generation stages, delineating potential failure modes, their underlying causes in threat models and system mechanisms, and their practical implications. Building on this analysis, we critically review 17 articles on privacy-preserving strategies for RAG systems. Our evaluation reveals critical gaps, including insufficient clinical validation, absence of standardized evaluation frameworks, and lack of automated assessment tools. We propose actionable directions based on these limitations and conclude with a call to action. This review provides researchers and practitioners with a structured framework for understanding privacy vulnerabilities in healthcare RAG and offers a roadmap toward developing systems that achieve both clinical effectiveness and robust privacy preservation.

arXiv:2511.11461v1 Announce Type: cross Abstract: Multi-step forecasting is often described through a simple rule of thumb: recursive strategies are said to have high bias and low variance, while direct strategies are said to have low bias and high variance. We revisit this belief by decomposing the expected multi-step forecast error into three parts: irreducible noise, a structural approximation gap, and an estimation-variance term. For linear predictors we show that the structural gap is identically zero for any dataset. For nonlinear predictors, however, the repeated composition used in recursion can increase model expressivity, making the structural gap depend on both the model and the data. We further show that the estimation variance of the recursive strategy at any horizon can be written as the one-step variance multiplied by a Jacobian-based amplification factor that measures how sensitive the composed predictor is to parameter error. This perspective explains when recursive forecasting may simultaneously have lower bias and higher variance than direct forecasting. Experiments with multilayer perceptrons on the ETTm1 dataset confirm these findings. The results offer practical guidance for choosing between recursive and direct strategies based on model nonlinearity and noise characteristics, rather than relying on traditional bias-variance intuition.

Managing and maintaining AI systems remains a challenge for many enterprises, particularly with the potential for agentic sprawl to expose businesses to risky entry points. Microsoft entered the observability fray with the launch of Agent 365 during its annual Ignite conference Tuesday. It described Agent 365 as the control plane for AI agents, serving as an observability layer for enterprises running any agent. The company said the platform “delivers unified observability” through telemetry, dashboards, and alerts to track every agent in use. Agent 365 supports any agents, whether built on Microsoft’s platforms or from third parties, including Adobe, Databricks, Cognition, and ServiceNow. “Agent 365 marks a new chapter in how organizations build, secure, and scale their agents. This is a shift from isolated experiments to enterprise readiness, where agents operate as part of a unified, governed, and productive system,” said Microsoft's president of business apps and agents, Charles Lamanna, in a blog post. What enterprises get Microsoft 365 has five capabilities: registry, access control, visualization, interoperability, and security. The first step in beginning observability tasks in Microsoft 365 is to log the agents that may be present, which will serve as a single source of truth. The company calls this registry Entra. “This single registry tracks the agents for every relevant role within your organization — IT, developers, security, and business leaders. And with the Agent Store, users can easily discover the right agents for their role and workflows directly within Microsoft 365 Copilot and Teams,” Microsoft said. Access Control would require agents to have a unique agent ID, allowing enterprise admins to limit access as needed. Organizations can set policies that agents must adhere to, and the tool can respond by blocking misbehaving agents. Agent 365 also offers a visual dashboard, allowing companies to see how their agents are connected, performance measurements, and task adherence. For enterprises already dealing with agent sprawl, Microsoft’s approach stands out because it bundles what’s usually a patchwork of add-on tools into a single, governed control plane. Most observability options today are either siloed features or separate platforms that add more complexity. Agent 365 treats agents as parts of the stack, giving IT and security teams unified oversight at a moment when that’s becoming essential . Companies like DataDog , Dynatrace and Splunk offer observability services for AI systems. The startup Chronosphere released capabilities similar to observability, focusing on debugging issues, and Raindrop also introduced its own observability tools for performance. Google also began offering an observability dashboard on its AI Agent Builder. “As agents multiply in numbers and sophistication, companies face a new kind of challenge: how to manage and govern agents responsibly and at scale, without rebuilding the trusted systems they rely on,” Lamanna said. “The clearest path forward is to manage agents the way you manage people, using the same infrastructure, apps, and protections that power your business today.”

Google launched yet another coding agent platform Tuesday, this time focused on developer teams collaborating to create agents that can execute complex tasks automatically, in a way that moves agents from being remotely controlled to actually independent. The platform, called Antigravity, is powered by Gemini 3 and is now available in public preview with “ generous rate limits on Gemini 3 Pro usage,” Google writes in a blog post accompanying the announcement. Antigravity is an agentic coding platform that aims to evolve the IDE toward an agent-first future with browser control capabilities, asynchronous interaction patterns, and an agent-first product design. Enterprises that are already bogged down by a growing volume of code to review, thanks in large part to the rise of AI code generation, are demanding more from asynchronous coding agents. They need asynchronous coding agents to help developers review coding projects, assess the elements, and perform tasks autonomously. For the public preview, Antigravity users can build agents using Gemini 3, Anthropic’s Sonnet 4.5 models, and OpenAI’s open-source gpt-oss . It will be compatible with developer environments running on major operating systems such as macOS, Linux and Windows. “We want Antigravity to be the home base for software development in the era of agents,” Google writes in the blog. “Our vision is to ultimately enable anyone with an idea to experience liftoff and build that idea into reality.” Google said it built Antigravity with four key tenets — trust, autonomy, feedback, and self-improvement — which it says sets it apart from other coding platforms because it focuses on a more collaborative development environment. Key tenets of development Enterprises today are either completely transparent about what's happening under the hood, or they don't show their work and simply split out code. The Antigravity team doesn't think either of these two extremes build trust. " Antigravity provides context on agentic work at a more natural task-level abstraction, with the necessary and sufficient set of artifacts and verification results, for the user to gain that trust. There is a concerted emphasis for the agent to thoroughly think through verification of its work, not just the work itself," according to Google. As for autonomy, Antigravity’s main interface, Editor View, mimics an IDE experience, standardizing what an agent might encounter while accomplishing its tasks. The agent is embedded in this interface so it can navigate it. However, Google plans to add “an agent-first Manager surface” that flips that idea around, meaning the interface is embedded into the agent. The Antigravity team built user feedback into “across every surface or artifact,” which will be automatically incorporated into agent execution. This would allow work to continue without requiring humans to stop the work to redirect the agent. With the human developer iterating with the agent, "self-improvement" becomes very essential. Its agent can tap a knowledge base to learn from past work or contribute new learnings. Google’s many coding agents Antigravity is not Google’s only coding platform; it’s not even its only coding agent with an IDE integration or asynchronous capabilities. It joins a long line of Google platforms aimed at helping developers work more efficiently. The coding assistant Jules is now integrated into IDEs , can be invoked via the CLI, and can also run asynchronously. Gemini CLI also works similarly. And there's Gemini Code Assist , which first launched last year. However, Antigravity will most likely have to compete more with coding agent platforms like Codex from OpenAI, Claude Code from Anthropic, and Cursor. Some people on X commented that Antigravity looks similar to Windsurf, which would make sense: Google hired the Windsurf team — including CEO Varun Mohan — in July and licensed the tech for $2.4 billion . Varun Mohan tweeted that this came from his team: So far, early Antigravity users have had mixed experiences, with many pointing to errors and slow generation. Editors note: This story was updated on November 18, 2025, to include more information.

arXiv:2505.10311v4 Announce Type: replace-cross Abstract: Conventional score-based diffusion models (DMs) may struggle with anisotropic Gaussian diffusion processes due to the required inversion of covariance matrices in the denoising score matching training objective \cite{vincent_connection_2011}. We propose Whitened Score (WS) diffusion models, a novel framework based on stochastic differential equations that learns the Whitened Score function instead of the standard score. This approach circumvents covariance inversion, extending score-based DMs by enabling stable training of DMs on arbitrary Gaussian forward noising processes. WS DMs establish equivalence with flow matching for arbitrary Gaussian noise, allow for tailored spectral inductive biases, and provide strong Bayesian priors for imaging inverse problems with structured noise. We experiment with a variety of computational imaging tasks using the CIFAR, CelebA ($64\times64$), and CelebA-HQ ($256\times256$) datasets and demonstrate that WS diffusion priors trained on anisotropic Gaussian noising processes consistently outperform conventional diffusion priors based on isotropic Gaussian noise. Our code is open-sourced at \href{https://github.com/jeffreyalido/wsdiffusion}{\texttt{github.com/jeffreyalido/wsdiffusion}}.

arXiv:2511.12690v1 Announce Type: new Abstract: Direct speech-to-speech translation (S2ST), in which all components are trained jointly, is an attractive alternative to cascaded systems because it offers a simpler pipeline and lower inference latency. However, direct S2ST models require large amounts of parallel speech data in the source and target languages, which are rarely available for low-resource languages such as Persian. This paper presents a direct S2ST system for translating Persian speech into English speech, as well as a pipeline for synthetic parallel Persian-English speech generation. The model comprises three components: (1) a conformer-based encoder, initialized from self-supervised pre-training, maps source speech to high-level acoustic representations; (2) a causal transformer decoder with relative position multi-head attention translates these representations into discrete target speech units; (3) a unit-based neural vocoder generates waveforms from the predicted discrete units. To mitigate the data scarcity problem, we construct a new Persian-English parallel speech corpus by translating Persian speech transcriptions into English using a large language model and then synthesizing the corresponding English speech with a state-of-the-art zero-shot text-to-speech system. The resulting corpus increases the amount of available parallel speech by roughly a factor of six. On the Persian-English portion of the CVSS corpus, the proposed model achieves improvement of 4.6 ASR BLEU with the synthetic data over direct baselines. These results indicate that combining self-supervised pre-training, discrete speech units, and synthetic parallel data is effective for improving direct S2ST in low-resource language pairs such as Persian-English