2608 papers
This paper proposes a training-free mathematical framework that selects the optimal machine learning model for SME cashflow forecasting by matching dataset attributes and user expertise against algorithmic capabilities, thereby eliminating the need for exhaustive model training runs.
This paper proposes a robust skin detection method using a two-stream U-Net that integrates color and texture information via a hierarchical attention mechanism, demonstrating improved accuracy and efficiency across challenging conditions and multiple datasets.
This study proposes an improved YOLOv8n-SDSC and DeepSORT framework integrated with an X-ray system to enhance the efficiency and accuracy of walnut detection and tracking for industrial sorting by incorporating Sea_Attention, Dysample, C2f_SCConv, and DIoU-based matching.
This paper introduces SOA-UNet, a lightweight deep learning framework that combines sparse feature encoding, Otsu-based attention guidance, and an Adaptive Boundary Loss to achieve accurate, fast, and interpretable brain tumor segmentation from MRI images.
This paper proposes a dual-modal machine learning framework that integrates textual (TF-IDF) and audio (MFCC) features to detect depression, achieving high accuracy (up to 99.38% with XGBoost on audio data) and demonstrating the potential of multimodal learning for scalable, non-invasive mental health diagnostics.
This paper proposes WMS-Net, a Wavelet-Mamba Synergistic Network that leverages multi-level Haar wavelet transforms, a Mamba-based Direction-Aware module, and a Cross-Task Attention Interaction mechanism to effectively address noise and feature entanglement, thereby achieving state-of-the-art joint semantic segmentation and height estimation from single RGB remote sensing images.
This paper proposes a lightweight, high-precision brain tumor detection model that combines super-resolution techniques with an artificial Transformer architecture featuring channel and pixel attention mechanisms, achieving 99.16% accuracy on a dataset of 7,023 MRI images while outperforming state-of-the-art deep learning models.
This paper introduces the DEDA-AFCQ algorithm, a novel edge detection enhancement method that utilizes coefficients from analytic functions within a q-Calculus framework to generate eight directional convolution masks, demonstrating superior structural fidelity and performance over existing techniques like CSKP and Mittag–Leffler based methods on standard datasets.
This paper proposes PaletteGAN, a palette-guided GAN framework that enhances structural preservation and color controllability in fashion sketch colorization by integrating brightness-preserving data augmentation, a perceptually optimized six-color palette construction, and a specialized generator-discriminator architecture, achieving superior performance in visual realism and design consistency on the Cleaned Maryland Dataset.
This paper proposes the Causal Resilience Graph Operator (CRGO), a unified dynamical system that integrates causal machine learning with graph learning to simultaneously model the evolution, shock propagation, and counterfactual resilience of global cross-border investment networks, enabling rigorous forecasting and systemic resilience auditing.