Robust Active Label Correction

Jan Kremer, Fei Sha, Christian Igel

Abstract: Active label correction addresses the problem of learning from input data for which noisy labels are available (e.g., from imprecise measurements or crowd-sourcing) and each true label can be obtai...