9189 papers
This paper presents findings from a large-scale global survey that explores diverse cultural perspectives on Generative AI, distilling community-defined understandings of culture to propose recommendations for more inclusive and sensitive AI development, including participatory approaches and frameworks for addressing cultural boundaries.
LTLGuard is a modular framework that enables resource-efficient open-weight language models (4B–14B parameters) to generate correct and conflict-free Linear Temporal Logic (LTL) specifications from informal requirements by combining constrained generation with lightweight symbolic reasoning for iterative consistency checking and refinement.
This paper challenges prior claims of Best-of-N's suboptimality by demonstrating that, under practical assumptions and when evaluated via win-rate rather than expected reward, properly tuned Best-of-N is both statistically and computationally optimal, while also proposing a simple variant that eliminates reward hacking without sacrificing performance.
This paper introduces TML-Bench, a benchmark for evaluating the end-to-end correctness and reliability of autonomous coding agents on Kaggle-style tabular machine learning tasks, demonstrating that the MiniMax-M2.1 model achieves the best aggregate performance across four competitions under varying time budgets.
This paper introduces SCORE, a novel model merging method that resolves singular subspace conflicts between domain-specific models by projecting them into a shared orthogonal basis, thereby significantly improving generalization to unseen domains compared to existing approaches.
This paper introduces SAHA, a novel jailbreak framework that targets deep safety attention heads in open-sourced large language models through ablation-impact ranking and layer-wise perturbation, achieving a 14% improvement in attack success rate over state-of-the-art baselines.
This paper proposes the Disentangled Safety Hypothesis (DSH), which reveals that large language models separate safety "recognition" and "refusal execution" into distinct geometric subspaces, enabling the development of the Refusal Erasure Attack (REA) to bypass safety mechanisms by surgically disabling the refusal axis while preserving harmful content generation.
This paper introduces PVminer, a benchmark for structured extraction of patient voice from patient-generated text, and presents PVminerLLM, a supervised fine-tuned large language model that significantly outperforms prompt-based baselines in extracting codes, sub-codes, and evidence spans to enable scalable analysis of non-clinical health drivers.
This study evaluates a dual-calibration algorithmic nudge and an LLM-based presentation nudge within a personalized diversity framework, finding that while algorithmic nudges effectively increase news consumption diversity and shift long-term reading habits toward balanced domestic and global coverage, LLM-based presentation nudges yield variable results and user-specific topic interest remains the strongest predictor of engagement.
The paper introduces BM25-V, a two-stage image retrieval framework that leverages sparse visual-word activations from a Sparse Auto-Encoder combined with Okapi BM25 scoring to achieve near-dense accuracy with high interpretability and computational efficiency.
This paper proposes "Proof-of-Guardrail," a system leveraging Trusted Execution Environments to provide cryptographic, verifiable attestation that AI agents enforce specific safety measures, thereby addressing the threat of falsely advertised safety while acknowledging remaining risks like active jailbreaking.
StreamWise is an adaptive, modular serving system that leverages heterogeneous hardware and dynamic resource management to enable cost-effective, high-quality real-time multi-modal generation (such as podcast videos) with sub-second startup delays, overcoming the latency and complexity challenges of coordinating diverse models at scale.
This paper introduces the concept of "ambiguity collapse," where large language models prematurely resolve genuinely ambiguous, value-laden terms into singular interpretations, and proposes a taxonomy of the resulting epistemic risks across process, output, and ecosystem levels alongside multi-layered mitigation strategies to preserve meaningful ambiguity.
This paper introduces Margin and Consistency Supervision (MaCS), an architecture-agnostic regularization framework that combines a hinge-squared margin penalty and a consistency regularizer to simultaneously enhance the calibration, robustness, and generalization of deep vision models without requiring additional data or architectural changes.
This paper presents Lexara, a user-centered toolkit designed to address the challenges of evaluating Large Language Models for Conversational Visual Analytics by providing real-world test cases, interpretable multi-format metrics, and an interactive interface that enables developers and end-users to assess model performance without programming expertise.
This paper presents a white-box analysis of the EleutherAI/gpt-j-6B model, demonstrating through contrastive prompting that its internal representation of trust aligns most closely with the Castelfranchi socio-cognitive model, thereby validating the feasibility of using LLM activation spaces to analyze socio-cognitive constructs and inform human-AI collaboration.
This paper proposes a deep ensemble learning framework that fuses four independent CNN-ViT hybrid models to overcome the performance bottlenecks of redundant feature representations, achieving state-of-the-art accuracy on remote sensing image classification datasets while maintaining computational efficiency.
The paper introduces MACRO, a self-evolving medical imaging agent that autonomously discovers and synthesizes reusable composite tools from verified execution trajectories to overcome the brittleness of static tool chains and enhance multi-step clinical decision-making across diverse domains.
This paper presents an international expert review on the clinical integration and translational readiness of emerging foundation and agentic AI in computational pathology, moving beyond technical performance to address the economic, regulatory, and operational barriers hindering real-world adoption in patient care.
This paper introduces JHCodec, a low-latency streaming neural audio codec that utilizes a self-supervised representation reconstruction (SSRR) loss to achieve state-of-the-art intelligibility and convergence speed without requiring additional lookahead or semantic encoder distillation.