2661 papers
This study demonstrates that a Graph Convolutional Neural Network can effectively estimate sixteen standard anthropometric measurements directly from 3D human body meshes, achieving promising accuracy on synthetic data while highlighting the challenges of domain adaptation when applied to real-world scans.
This paper presents an auditable, edge-deployable AI safety architecture that combines conformal prediction with input quality assurance to eliminate autonomous false-negative tuberculosis discharges across diverse, resource-constrained international cohorts while maintaining high diagnostic accuracy and reducing unnecessary confirmatory testing.
This systematic literature review analyzes sixteen studies on adaptive context reliance control in noisy Retrieval-Augmented Generation systems, identifying critical gaps in current noise-aware memorization strategies and proposing the ACRC framework to dynamically balance an LLM's internal knowledge against retrieved evidence quality.
The paper proposes SET-CNN, a novel framework that integrates DenseNet121, ResNet50, and VGG16 through feature-level fusion and triplet semi-hard loss optimization to generate robust image embeddings, achieving a 7.6% improvement in retrieval performance over individual models on the CIFAR-10 dataset.
This paper proposes and validates a hybrid Computational Analytics model (HCAM) that integrates data mining techniques for feature structuring with supervised machine learning classifiers, demonstrating statistically significant improvements in predictive accuracy and interpretability over standalone methods on the UCI Heart Disease dataset.
This paper proposes RLo-RAG, a Reinforcement Learning-based Multi-Round Retrieval-Augmented Generation model enhanced with LoRA, which utilizes a dynamic reward function to iteratively retrieve high-relevance document chunks and significantly improves retrieval precision and generation quality for industrial documents compared to traditional RAG methods.
This paper proposes MDFR-Net, a novel deep learning framework that enhances aerial small object detection by integrating Multi-scale Hierarchical Attentional Fusion, Orientation Decoupled Feature Enhancer, and Hierarchical Convolution Pooling Fusion modules to effectively address challenges related to minimal scale, diverse orientations, and complex background interference.
This paper presents a machine learning-based hybrid task scheduler that optimizes resource allocation between classical and quantum processors using real graph-derived workloads, demonstrating superior performance in completion time, makespan, and throughput compared to traditional and rule-based baselines.
This paper proposes DA-LightNet, a lightweight, unified deep learning network that effectively removes rain and haze in real-time on resource-constrained edge devices by employing a degradation-aware modulation module and efficient convolutional blocks to balance high restoration quality with low computational cost.
This paper proposes a context-adaptive framework for multi-turn emotional speech synthesis that leverages historical dialogue memory and a bimodal joint input of speech signals and text to dynamically predict and control acoustic features, thereby overcoming the limitations of traditional transcription-based methods to achieve natural and emotionally coherent responses.