Deep k-NN for Noisy Labels

Dara Bahri · Heinrich Jiang · Maya Gupta

Keywords: [ Information Theory and Estimation ] [ Learning Theory ] [ Supervised Learning ] [ Other ]

[ Abstract ]
Tue 14 Jul 9 a.m. PDT — 9:45 a.m. PDT
Wed 15 Jul 8 p.m. PDT — 8:45 p.m. PDT

Abstract: Modern machine learning models are often trained on examples with noisy labels that hurt performance and are hard to identify. In this paper, we provide an empirical study showing that a simple $k$-nearest neighbor-based filtering approach on the logit layer of a preliminary model can remove mislabeled training data and produce more accurate models than many recently proposed methods. We also provide new statistical guarantees into its efficacy.

Chat is not available.