2731 papers
This paper introduces boundary-damped wave equilibrium propagation, a physical learning architecture that combines conservative interior dynamics with localized boundary dissipation to achieve finite-time gradient convergence, spectral stability, and high accuracy on tasks like MNIST while maintaining strict energy-balance and gradient-error bounds.
AiWebArmor is a novel, high-throughput AI framework that combines ModernBERT-based semantic analysis with statistical features via an explainable tree-boosting classifier to achieve 99.92% multi-class threat detection accuracy and a 0.01% false-positive rate with sub-millisecond latency in production web environments.
This paper introduces a unified generative framework that leverages a single swarm intelligence simulation, governed by fundamental interaction rules, to simultaneously produce diverse outputs across architecture, visual art, music, and narrative domains through specialized domain-specific mappings.
This paper proposes a fairness-constrained credit scoring framework using mobile money data that significantly reduces gender-based approval gaps and regulatory violations while maintaining high accuracy, though intersectional analysis reveals that such constraints alone cannot fully resolve compound socioeconomic disparities affecting low-income women.
The paper introduces CheXficient, a chest X-ray foundation model that leverages representation-guided selective data curation to achieve competitive performance across diverse clinical tasks using only a fraction of the data and computational resources required by traditional large-scale pretraining approaches.
This paper introduces a distribution-conditional benchmark demonstrating that feature-attribution explanations for network intrusion alerts can either improve, degrade, or leave unchanged the utility of alert queues depending on specific threat distributions and detector characteristics, thereby supporting the use of locally validated, explanation-aware re-ranking as a complementary strategy to score-only ordering.
This paper presents enhanced experimental validation demonstrating that the Query-Adaptive Coordinate Ordering method achieves an average 2.84× speedup in exact k-NN search across high-dimensional datasets while maintaining perfect recall, with performance gains primarily driven by feature correlation rather than nominal dimensionality.
This study evaluates the predictive performance and cross-method explanation consistency of an XGBoost classifier on a heart-disease dataset, demonstrating that while the model achieves strong accuracy and moderate-to-strong agreement among global explanation methods (native importance, permutation importance, and SHAP), these dimensions are distinct and do not inherently guarantee clinical validity.
This paper introduces AmbiGraph-Coref, a graph-based model enhanced with confidence calibration and an explicit unresolvable mechanism that achieves high performance in resolving ambiguous coreference in synthetic social media text, while emphasizing the need for future validation on authentic real-world data.
This study leverages a large-scale dataset and a quality-aware machine learning model to accurately distinguish playback-quality issues from low user interest in rapid video swipes, achieving high predictive performance and enabling targeted optimization of streaming experiences.