2661 papers
This paper proposes and evaluates a perturbation-based audit and mitigation framework that demonstrates how entity noise increases high-confidence false relation risks in LLM-generated scientometric graphs, revealing that entity- and evidence-aware abstention strategies effectively outperform simple confidence filtering in preserving graph integrity.
WSL-Det is a real-time object detector built on the Vajra architecture that integrates wave-based attention for frequency-aware global interaction and sequential orthogonal strip convolutions for anisotropic localization, achieving state-of-the-art accuracy and speed on industrial strip defect datasets like leather and NEU-DET.
This study developed and validated an explainable, context-aware machine learning framework using electronic medical record data from a Ghanaian hospital that successfully predicts medication adherence among chronic disease patients with high accuracy (89.5%) and identified a compact, resource-efficient feature set suitable for deployment in low-resource health systems.
The paper introduces Chest2Vec, a multipurpose, instruction-conditioned text encoder available in 0.6B and 4B parameter sizes that effectively processes both chest radiograph and CT reports for tasks like retrieval, classification, and image-text supervision, demonstrating superior performance on CT reports compared to existing models.
This paper proposes Instance-Aware Partial Negative Completion (IPNC), a structure-regularized framework that learns bounded soft negative targets by coupling a multi-output predictor with feature and score-based graph structures to effectively handle ambiguous candidate labels in partial multi-label learning, achieving superior empirical performance across diverse datasets.
The paper proposes FCA-Attention, a neuro-symbolic framework that integrates Formal Concept Analysis-derived lattice priors with BERT representations to enhance joint clinical entity and relation extraction, achieving improved performance and interpretability without relying on external knowledge graphs.
This paper proposes the Heterogeneity-Adaptive Asynchronous ADMM (HA-ADMM) algorithm, which utilizes -weighted push-sum mixing and adaptive penalty updates to achieve near-stationarity in heterogeneous nonconvex multi-agent optimization by explicitly characterizing and mitigating the effects of gradient dissimilarity, Lipschitz spread, and communication delays.
This paper demonstrates that an apparent key-dependent timing side-channel in Bouncy Castle's ML-DSA-65 implementation is actually a statistical artifact of natural rejection-sampling variation, which vanishes when timing is analyzed within a pre-registered protocol that conditions on the exact number of iterations performed, thereby confirming the implementation's security across all standardized parameter sets.
This paper investigates hallucinations in prominent AI chatbots like GPT-4, Gemini, Claude, and Copilot through literature review and surveys, highlighting their tendency to generate fabricated information and citation errors in academic contexts while proposing mitigation strategies such as manual review and fact-checking.
This paper proposes a hierarchical framework that leverages a compact large language model for semantic task planning and decomposition while delegating resource allocation and execution decisions to a constrained optimization layer, thereby bridging natural-language specifications with mathematically rigorous multidisciplinary optimization for Edge IoT systems.