2661 papers
This paper proposes a novel Density-Adaptive model based on Motif Matrix (NDAMM) that leverages multi-scale spatial feature fusion and atrous pyramidal temporal pooling to achieve accurate, interpretable, and efficient automated detection of premature atrial and ventricular contractions, outperforming existing methods across public databases.
This paper proposes and evaluates a lightweight, probe-based policy for selective classification on OpenML tabular tasks that dynamically chooses between raw predictions, temperature scaling, and Dirichlet calibration, demonstrating that such conditional selection improves both probability accuracy and abstention quality compared to universal calibration or no calibration.
Āyurpadaviśleṣikā is an AI-assisted digital platform that organizes over 4,000 medicinal entries from classical Sanskrit texts using the five-fold Ayurvedic framework to address knowledge fragmentation and enhance accessibility for diverse users through natural language processing, computer vision, and expert curation.
This paper proposes a feature-based, interpretable machine learning approach using temporal and structural legislative history that significantly outperforms existing baselines in predicting proposition-level approval outcomes in the Brazilian Chamber of Deputies, offering practical tools for enhanced legislative monitoring and transparency.
This paper proposes a cross-dataset framework that integrates evidence-filtered SHAP attributions with pedagogically grounded LLM feedback to generate actionable, scenario-compatible explanations for student-risk prediction across diverse educational contexts, demonstrating improved compliance and robustness through validation on both cross-sectional and longitudinal datasets.
The paper introduces EEG-LM, a foundation model that bridges brain signals and natural language by aligning EEG representations with large language model embeddings through diffusion-augmented contrastive learning, thereby achieving state-of-the-art performance, strong generalization, and interpretable emotion recognition across multiple datasets.
This paper proposes E-GCNet, a hybrid deep learning framework that integrates Adaptive Min-Max Vector Normalization, an Improved Weighted Wrapper-Filter for feature selection, and a soft-voting ensemble of Enhanced GoogleNet and Capsule Networks to achieve highly accurate and reliable cognitive state detection with 97.4% accuracy.
This systematic review, conducted according to PRISMA 2020 guidelines, evaluates the potential of various deep learning algorithms for the early detection of chronic diseases while highlighting current limitations such as data scarcity and interpretability, and proposing future directions to develop more reliable and clinically beneficial diagnostic systems.
This study presents a configurable LLM system utilizing a Minimal Atomic Unit (MAU) framework that achieves high-precision, source-traceable data extraction for meta-analyses, significantly outperforming non-decomposed approaches while maintaining a negligible hallucination rate across diverse clinical domains.
This paper demonstrates that integrating Tensor-Train layers into Physics-Informed Neural Networks significantly reduces parameter counts by up to 8.8× while maintaining 97% of the original accuracy, with compression efficiency heavily dependent on the regularity of the underlying PDE solution.