2787 papers
This study proposes a multimodal deep learning framework that integrates chest X-ray images with demographic metadata using a ConvNeXt Large backbone, achieving a 96.88% macro-average AUC on the NIH ChestX-ray14 dataset and demonstrating superior performance and interpretability compared to existing image-only models.
This paper proposes HRDAWNet, a novel attention framework integrating Triple-Factorized Attention and Dynamic Adaptive Windowing to achieve state-of-the-art accuracy in classifying premature atrial and ventricular contractions by effectively capturing complex spatio-temporal dependencies and adapting to dynamic contextual information.
This paper presents a multi-semester case study demonstrating that replacing traditional XAMPP-based PHP development environments with Docker and DevContainers in Bahraini universities reduces student setup time by over 90% and significantly lowers extraneous cognitive load, thereby preventing configuration errors from jeopardizing the entire course.
This paper proposes a learning-based heuristic for dynamic path planning that integrates a hop-aware graph neural network (HopGNN) with a convolutional gated recurrent unit (ConvGRU) to effectively capture multi-scale spatial topologies and temporal environmental changes, demonstrating superior success rates and search efficiency over existing methods like GCN-A*, GAT-A*, and D* Lite in simulated grid environments.
This paper presents a multi-domain benchmark evaluating five large language models, a rule-based lexicon, and a supervised transformer against human annotations, revealing that while top-tier models like Claude Opus 4.7 and GPT-5.5 achieve strong alignment with humans, significant disagreement persists across models and with human labels—particularly on informal text—demonstrating that model selection materially impacts sentiment classification outcomes.
This paper proposes a novel, model-agnostic AST-level watermarking framework that embeds cryptographically verifiable signatures into the syntactic topology of AI-generated code through semantics-preserving structural mutations, achieving robust provenance attribution and significantly outperforming existing text-level methods against common code transformations like formatting, renaming, and dead-code injection.
This systematic review of 297 studies reveals that while SHAP and LIME are widely adopted for explainability in financial machine learning—particularly for credit modeling—they suffer from a critical lack of formal evaluation, with only 9.1% of papers assessing explanation quality, highlighting an urgent need for stronger validation and deployment standards to ensure regulatory compliance and trust.
This paper proposes a secure software over-the-air (SOTA) update framework that enhances the security of OTA transmissions by encrypting software binaries with AES-256-GCM and embedding them into grayscale images using chaotic-map-guided LSB steganography to obscure the payload and prevent tampering.
This study demonstrates that while generative AI significantly enhances vocabulary acquisition in EFL learners compared to conventional instruction, linear regression models relying primarily on linguistic indicators outperform complex machine learning algorithms and behavioral metrics in predicting these learning gains.
This paper proposes a quantum-resilient, identity-aware communication architecture for edge IoT that utilizes frozen DINOv2 visual encoders combined with Siamese and Graph Neural Networks to verify participants via quantized embeddings, ensuring secure, low-latency, and bandwidth-efficient sessions for critical applications like telepresence and remote healthcare.