Training Noise-Robust Deep Neural Networks via Meta-Learning
Authors:
- 1 June 2020
- conference paper
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- p. 4523-4532
Abstract
Label noise may significantly degrade the performance of Deep Neural Networks (DNNs). To train noise-robust DNNs, Loss correction (LC) approaches have been introduced. LC approaches assume the noisy labels are corrupted from clean (ground-truth) labels by an unknown noise transition matrix T. The backbone DNNs and T can be trained separately, where T is approximated with prior knowledge. For example, T is constructed by stacking the maximum or mean predic- tions of the samples from each class. In this work, we pro- pose a new loss correction approach, named as Meta Loss Correction (MLC), to directly learn T from data via the meta-learning framework. The MLC is model-agnostic and learns T from data rather than heuristically approximates it using prior knowledge. Extensive evaluations are conducted on computer vision (MNIST, CIFAR-10, CIFAR-100, Cloth- ing1M) and natural language processing (Twitter) datasets. The experimental results show that MLC achieves very com- petitive performance against state-of-the-art approaches.Keywords
This publication has 21 references indexed in Scilit:
- ImageNet classification with deep convolutional neural networksCommunications of the ACM, 2017
- DeepLab: Semantic Image Segmentation with Deep Convolutional Nets, Atrous Convolution, and Fully Connected CRFsIEEE Transactions on Pattern Analysis and Machine Intelligence, 2017
- Learning Deep Networks from Noisy Labels with Dropout Regularization2016
- Co-Saliency Detection via a Self-Paced Multiple-Instance Learning FrameworkIEEE Transactions on Pattern Analysis and Machine Intelligence, 2016
- Training deep neural-networks based on unreliable labels2016
- When Face Recognition Meets with Deep Learning: An Evaluation of Convolutional Neural Networks for Face Recognition2015
- Classification with Noisy Labels by Importance ReweightingIEEE Transactions on Pattern Analysis and Machine Intelligence, 2015
- Making risk minimization tolerant to label noiseNeurocomputing, 2015
- Rich Feature Hierarchies for Accurate Object Detection and Semantic Segmentation2014
- Gradient-based learning applied to document recognitionProceedings of the IEEE, 1998