<ul><li>A new semi-supervised deep learning framework has been proposed to enhance classification performance by prioritizing critical training samples based on their proximity to test data.</li><li>The method aims to improve model generalization and robustness, especially in scenarios with noisy or imbalanced datasets, by focusing on the most informative examples and utilizing distance-based sample weights.</li><li>The approach addresses challenges of limited labeled data while ensuring scalability by incorporating techniques like uncertainty consistency and graph-based representations.</li><li>Experiments on twelve benchmark datasets have shown significant enhancements in accuracy, precision, and recall, surpassing existing methods and offering a practical solution for semi-supervised learning in fields like healthcare and security.</li></ul>

Enhancing Classification with Semi-Supervised Deep Learning Using Distance-Based Sample Weights

Discover more