SIGUA: Forgetting may make learning with noisy labels more robust

Bo Han*, Gang Niu, Xingrui Yu, Quanming Yao, Miao Xu, Ivor W. Tsang, Masashi Sugiyama

*Corresponding author for this work

Research output: Chapter in book/report/conference proceedingConference proceedingpeer-review

56 Citations (Scopus)


Given data with noisy labels, over-parameterized deep networks can gradually memorize the data, and fit everything in the end. Although equipped with corrections for noisy labels, many learning methods in this area still suffer overfitting due to undesired memorization. In this paper, to relieve this issue, we propose stochastic integrated gradient underweighted ascent (SIGUA): in a minibatch, we adopt gradient descent on good data as usual, and learning-rate-reduced gradient ascent on bad data; the proposal is a versatile approach where data goodness or badness is w.r.t. desired or undesired memorization given a base learning method. Technically, SIGUA pulls optimization back for generalization when their goals conflict with each other; philosophically, SIGUA shows forgetting undesired memorization can reinforce desired memorization. Experiments demonstrate that SIGUA successfully robustifies two typical base learning methods, so that their performance is often significantly improved.

Original languageEnglish
Title of host publicationProceedings of the 37th International Conference on Machine Learning, ICML 2020
EditorsHal Daumé III, Aarti Singh
PublisherML Research Press
Number of pages11
ISBN (Electronic)9781713821120
Publication statusPublished - Jul 2020
Event37th International Conference on Machine Learning, ICML 2020 - Virtual, Online
Duration: 13 Jul 202018 Jul 2020

Publication series

NameProceedings of Machine Learning Research
ISSN (Print)2640-3498


Conference37th International Conference on Machine Learning, ICML 2020
Internet address

Scopus Subject Areas

  • Computational Theory and Mathematics
  • Human-Computer Interaction
  • Software


Dive into the research topics of 'SIGUA: Forgetting may make learning with noisy labels more robust'. Together they form a unique fingerprint.

Cite this