Label noise
Confident learning: calibrates probabilities per class, estimates the share of wrong labels and lists the most suspicious rows with a label quality score.
Confident learning: calibrates probabilities per class, estimates the share of wrong labels and lists the most suspicious rows with a label quality score.