Deep k-NN for Noisy Labels

2020/04/26 by Bahri, Dara, Jiang, Heinrich, Gupta, Maya · 1 citation
#FOS: Computer and information sciences #Machine Learning (cs.LG) #Machine Learning (stat.ML)

paper · doi:10.48550/arxiv.2004.12289

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.

Cited by

• Detecting and Rectifying Noisy Labels: A Similarity-based Approach

Related

• kNet: A Deep kNN Network To Handle Label Noise
• KNN-enhanced Deep Learning Against Noisy Labels
• The Resistance to Label Noise in K-NN and DNN Depends on its\n Concentration
• Learning Deep Networks from Noisy Labels with Dropout Regularization
• An Embedding is Worth a Thousand Noisy Labels