3031 papers
This paper proposes HyperTuneMed, an intelligent framework that utilizes a hyper-parameterized logistic regression algorithm to predict diseases based on user symptoms with superior accuracy (up to 98.983%) and provide personalized preventive healthcare recommendations.
This paper demonstrates that while automated tools like Slither and Mythril effectively detect known vulnerability patterns, formal specification is essential for verifying logical correctness and ensuring safety in blockchain-based EEG data governance, as it uniquely identified a seeded array out-of-bounds vulnerability that automated tools missed.
This paper proposes an innovative traffic sign prediction framework that integrates Non-local Means preprocessing, a modified U-Net architecture, Dense CNN classification, and Rock Hyrax Swarm Optimization for hyperparameter tuning, achieving state-of-the-art accuracy of 99.64% on the TT100K and 99.6% on the GTSDB datasets.
This paper introduces EDNet-20, a novel deep learning framework optimized with photometric data augmentation and explainable AI techniques that achieves superior multi-label ocular disease detection accuracy compared to state-of-the-art models, offering a promising tool for early clinical diagnosis.
This paper introduces Empathic Extended Prompting, a real-time framework that integrates facial affective descriptors into Large Language Model interactions to enhance perceived emotional intelligence and mood responsiveness without compromising usability or triggering uncanny perceptions.
This paper evaluates a modular LLM-based multi-agent system for manufacturing information access, finding that while top-tier models outperform average human users in accuracy and speed on traceability tasks under specific data conditions, the system's effectiveness remains conditional on model selection, data quality, and prompt design, necessitating continued human oversight.
This paper presents a lightweight, multi-scale feature extraction network that enables real-time, high-accuracy hand gesture control for resource-constrained UAVs by employing a three-stage downsampling architecture and a hybrid module combining parallel dilated convolutions with a selective state space model.
This study proposes the Maritime Anomaly Detection Quality Index (MADQI), a novel composite evaluation framework that addresses the lack of labelled data in unsupervised AIS anomaly detection by integrating geospatial analysis with four specific metrics to robustly assess model performance for enhanced ocean monitoring and sustainable governance.
This paper proposes ADDETECTOR, a secure and accurate Alzheimer's disease detection system for intelligent elderly care that integrates topic-based language feature enhancement with a dynamic differential privacy mechanism to overcome low accuracy and privacy leakage challenges while demonstrating superior performance on the ADReSS dataset.
This paper presents an AI-driven approach for the automated 3D segmentation and visualization of 11 anatomical structures in the lower forearm, aiming to enhance musculoskeletal ultrasound proficiency and serve as a bedside teaching tool.