RSTLess-research/NCOD-Learning-with-noisy-labels

We propose a new method to train machine learning models in the presence of noisy labels. Our approach combines distance to class centroids with discounting to reduce outlier samples. This method effectively addresses the issue of noisy labels.

Python