Active label cleaning: Improving dataset quality under resource constraints.
Mélanie Bernhardt, Daniel C. Castro, Ryutaro Tanno, Anton Schwaighofer, Kerem Can Tezcan, Miguel Monteiro, Shruthi Bannur, Matthew P. Lungren, Aditya V. Nori, Ben Glocker, Javier Alvarez-Valle, Ozan Oktay · arXiv (Cornell University) · 2021
Imperfections in data annotation, known as label noise, are detrimental to the training of machine learning models and have an often-overlooked confounding effect on the assessment of model performance. Nevertheless, employing experts to remove label noise by fully re-annotating large datasets is infeasible in resource-constrained settings, such as healthcare. This work advocates for a data-driven approach to prioritising samples for re-annotation - which we term label cleaning. We propose to rank instances according to estimated label correctness and labelling difficulty of each sample, and introduce a simulation framework to evaluate relabelling efficacy. Our experiments on natural images and on a new medical imaging benchmark show that cleaning noisy labels mitigates their negative impact on model training, evaluation, and selection. Crucially, the proposed active label cleaning enables correcting labels up to 4 times more effectively than typical random selection in realistic conditions, making better use of experts' valuable time for improving dataset quality.