Probabilistic End-to-end Noise Correction for Learning with Noisy Labels

TL;DR

This paper introduces PENCIL, an end-to-end framework that effectively corrects noisy labels during training, improving deep learning performance without requiring clean datasets or prior noise information.

Contribution

PENCIL is a novel, flexible, and robust method that jointly updates network parameters and label distributions, outperforming existing approaches on various noisy datasets.

Findings

PENCIL outperforms state-of-the-art methods on synthetic and real-world noisy datasets.

It is robust even on clean datasets.

The method is independent of network architecture and does not need auxiliary data.

Abstract

Deep learning has achieved excellent performance in various computer vision tasks, but requires a lot of training examples with clean labels. It is easy to collect a dataset with noisy labels, but such noise makes networks overfit seriously and accuracies drop dramatically. To address this problem, we propose an end-to-end framework called PENCIL, which can update both network parameters and label estimations as label distributions. PENCIL is independent of the backbone network structure and does not need an auxiliary clean dataset or prior information about noise, thus it is more general and robust than existing methods and is easy to apply. PENCIL outperforms previous state-of-the-art methods by large margins on both synthetic and real-world datasets with different noise types and noise rates. Experiments show that PENCIL is robust on clean datasets, too.