3031 papers
This paper proposes a unified hybrid framework that integrates the interpretability and efficiency of classical Digital Signal Processing with the robustness and adaptability of Machine Learning to effectively address speech enhancement and image restoration challenges under complex, non-stationary degradation conditions.
This Multivocal Literature Review synthesizes evidence from 24 sources to conclude that while Large Language Models show promise for specific, code-centric Quantum Software Engineering tasks like synthesis and repair, they currently function primarily as bounded assistive tools rather than robust, lifecycle-spanning agents due to significant limitations in semantic accuracy, backend execution, and domain coverage.
This study leverages machine learning and SHAP-based explainability to effectively detect fraudulent online job postings in Ghana, identifying Multinomial Naive Bayes and Logistic Regression as top-performing models and highlighting educational requirements, application deadlines, and job categorization as key fraud indicators.
This paper proposes a context-enriched spatio-temporal graph neural network framework that integrates heterogeneous urban data to significantly improve electric vehicle charging station energy demand forecasting accuracy compared to traditional temporal models across diverse urban and semi-rural settings.
This paper presents a robust hybrid deepfake detection framework that integrates ResNet-50, XceptionNet, and Vision Transformer for spatial feature extraction with Bidirectional LSTM for temporal modeling, achieving state-of-the-art performance with 91.07% accuracy on the DFD and FF++ datasets.
This paper presents HMFuzzer, a multi-agent cooperative fuzzing framework that enhances HarmonyOS API vulnerability discovery by leveraging Android code retrieval, multimodal document interpretation, reinforcement learning for prompt optimization, and automated crash analysis to overcome the scarcity and low quality of existing ecosystem documentation.
This paper presents WhispiTree, a multi-sensory conversational AI system grounded in the "tree hole" metaphor and CBT principles that, through a nine-week randomized controlled trial, demonstrated significant efficacy in reducing anxiety and depression among college students while offering superior accessibility, privacy, and real-time feedback compared to traditional counseling.
This paper presents a systematic pipeline for robust masked face recognition that combines a fine-tuned YOLOv8 detector with landmark-guided upper-face cropping, demonstrating through extensive benchmarking that this multi-stage optimization significantly outperforms single-model approaches, particularly when using FaceNet with cosine similarity.
This paper proposes and evaluates two Integer Linear Programming (ILP) approaches for orchestrating the activation and charging cycles of energy-neutral sensors to ensure continuous temporal and spatial coverage in outdoor monitoring networks, analyzing the trade-offs between implementation complexity, solving time, and solution optimality.
This paper proposes the Dynamic Explainable Framework (DEF), a three-stage pipeline integrating time-series Transformers, Graph Convolutional Networks, and post-hoc explainability techniques to achieve high-accuracy intrusion detection, significantly reduce analyst alert fatigue by prioritizing critical incidents, and provide transparent, near-real-time rationales for Security Operation Center environments.