Noise-Tolerant Paradigm for Training Face Recognition CNNs

Wei Hu, Yangyu Huang, Fan Zhang, Ruirui Li; The IEEE Conference on Computer Vision and Pattern Recognition (CVPR), 2019, pp. 11887-11896


Benefit from large-scale training datasets, deep Convolutional Neural Networks(CNNs) have achieved impressive results in face recognition(FR). However, tremendous scale of datasets inevitably lead to noisy data, which obviously reduce the performance of the trained CNN models. Kicking out wrong labels from large-scale FR datasets is still very expensive, although some cleaning approaches are proposed. According to the analysis of the whole process of training CNN models supervised by angular margin based loss(AM-Loss) functions, we find that the  distribution of training samples implicitly reflects their probability of being clean. Thus, we propose a novel training paradigm that employs the idea of weighting samples based on the above probability. Without any prior knowledge of noise, we can train high performance CNN models with largescale FR datasets. Experiments demonstrate the effectiveness of our training paradigm. The codes are available at

Related Material

author = {Hu, Wei and Huang, Yangyu and Zhang, Fan and Li, Ruirui},
title = {Noise-Tolerant Paradigm for Training Face Recognition CNNs},
booktitle = {The IEEE Conference on Computer Vision and Pattern Recognition (CVPR)},
month = {June},
year = {2019}