Tag
The paper proposes UASPL, a method that integrates predictive reliability into sample selection for self-paced learning using evidential neural networks, improving classification performance and interpretability.