Prediction Consistency Regularization for Learning with Noise Labels Based on Contrastive Clustering.

In recent years, a notable trend has emerged in the form of contrastive-learning-based methodologies tailored to address the challenges posed by noisy labels.

also in 1,958 other papers