6225 papers
STAIRS-Former is a novel transformer architecture for offline multi-task multi-agent reinforcement learning that leverages spatio-temporal attention, an interleaved recursive structure, and token dropout to effectively handle varying agent populations and long-horizon dependencies, achieving state-of-the-art performance across diverse benchmarks.
This paper introduces OSCBench, a novel benchmark derived from instructional cooking data that evaluates the ability of text-to-video models to generate accurate and temporally consistent object state changes, revealing that current models struggle significantly with this capability despite their progress in other areas.
This paper introduces the "Agent Scaling Law" and the AgentProfile framework to demonstrate that the capabilities of educational AI agents scale predictably with structured profile richness—specifically role clarity, skill depth, tool completeness, runtime capability, and educator expertise—rather than solely through increased model size, as validated by the EduClaw platform's deployment of over 330 agent profiles.
This paper introduces a new dataset and demonstrates that surface electromyography (sEMG) signals from facial and neck muscles can reliably decode affective states, particularly frustration, during both phonated and silent speech, highlighting their potential for affect-aware silent speech interfaces.
This paper proposes an architecture for an "Agentic Operating System for Hospital" that adapts the OpenClaw framework to safely deploy LLM agents in clinical environments by integrating a restricted execution environment, document-centric interactions, page-indexed long-term memory, and a curated medical skills library to ensure reliability, security, and auditability in dynamic workflows.
This paper introduces Transfer Aware Dijkstra (TAD), an optimized algorithm that outperforms the state-of-the-art RAPTOR-based MR method in public transit routing by correctly handling buffer times through trip-sequence scanning while maintaining optimal results and achieving over a two-fold speed-up.
This study evaluates the potential for gender bias in generative AI-assisted recruitment by analyzing how a state-of-the-art model (GPT-5) suggests occupations for simulated Italian graduates, revealing that while job recommendations remain neutral, the model perpetuates gender stereotypes by attributing emotional traits to women and analytical traits to men.
The paper introduces CINDI, an unsupervised probabilistic framework based on conditional normalizing flows that unifies anomaly detection and imputation into a single end-to-end system to effectively restore data integrity in noisy multivariate time series, such as those found in power grids.
This paper proposes the Compression-Consistency Principle to demonstrate that language models' apparent preference for truth is not an intrinsic drive but a byproduct of compression pressure favoring hypotheses that offer shorter, more internally consistent descriptions of the training data.
This paper proposes an anomaly detection framework for multivariate time-series that leverages conditional normalizing flows with explicit inductive biases to constrain latent representations to prescribed temporal dynamics, thereby defining anomalies as violations of these dynamics rather than low observation likelihoods.
This paper proposes a free energy-based social bandit learning algorithm that enables agents to effectively identify and leverage the behavioral information of both expert and non-expert peers without reward observations, thereby achieving optimal policy convergence and significantly enhanced learning performance with logarithmic regret.
This paper analyzes the semantics and usage of Wikidata qualifiers to develop a refined taxonomy based on frequency and diversity metrics, aiming to improve knowledge graph querying, inference, and contributor guidance.
This paper introduces the Stability and Safety-Governed Memory (SSGM) framework to address critical risks like memory corruption, semantic drift, and privacy vulnerabilities in evolving LLM agents by decoupling memory evolution from execution through consistency verification, temporal decay modeling, and dynamic access control.
This paper presents NETHIC, an automatic text classification tool that combines scalable neural networks with hierarchical taxonomies and document embeddings to achieve significant improvements in both effectiveness and efficiency across generic and domain-specific corpora.
This paper introduces Deliberative Collective Intelligence (DCI), a structured multi-agent framework that employs typed epistemic acts and a convergent flow algorithm to significantly improve performance on complex, non-routine reasoning tasks through accountable deliberation, despite incurring high computational costs and underperforming on routine decisions.
HELM is a novel framework for hierarchical multi-label image classification in remote sensing that combines hierarchy-specific class tokens, graph-based structural encoding, and self-supervised learning to achieve state-of-the-art performance, particularly in low-label scenarios.
This paper proposes a mechanistic fairness audit framework to localize demographic bias within individual attention heads of CLIP's vision encoder, demonstrating that ablating specific heads can reduce gender bias while improving accuracy, whereas age bias appears more diffusely encoded.
DocSage is an end-to-end agentic framework that addresses the limitations of existing RAG systems in multi-document, multi-entity question answering by integrating dynamic schema discovery, error-aware structured extraction, and schema-aware relational reasoning to significantly improve cross-document evidence aggregation and accuracy.
This paper introduces the SDec-POMDP framework and the exact RS-SDA* algorithm to unify decentralized and multiagent POMDPs under a semi-decentralized approach that models communication uncertainty through time-distributed action and observation histories.
This paper presents a framework for automating the acquisition of specialized procedural agent skills by systematically mining open-source repositories to extract, standardize, and evaluate capabilities like mathematical visualization, demonstrating that such methods can significantly enhance LLM performance in autonomous workflows without requiring model retraining.