Identifying Mislabeled Data using the Area Under the Margin Ranking

Abstract

Not all data in a typical training set help with generalization; some samples can be overly ambiguous or outrightly mislabeled. This paper introduces a new method to identify such samples and mitigate their impact when training neural networks. At the heart of our algorithm is the Area Under the Margin (AUM) statistic, which exploits differences in the training dynamics of clean and mislabeled samples. A simple procedure - adding an extra class populated with purposefully mislabeled threshold samples - learns a AUM upper bound that isolates mislabeled data. This approach consistently improves upon prior work on synthetic and real-world datasets. On the WebVision50 classification task our method removes 17% of training data, yielding a 1.6% (absolute) improvement in test error. On CIFAR100 removing 13% of the data leads to a 1.2% drop in error.

arXiv commentNeurIPS 2020

References in corpus (18)

• PyTorch: An Imperative Style, High-Performance Deep Learning Library
• Learning to Reweight Examples for Robust Deep Learning
• Training Deep Neural Networks on Noisy Labels with Bootstrapping
• DivideMix: Learning with Noisy Labels as Semi-supervised Learning
• Do ImageNet Classifiers Generalize to ImageNet?
• A Closer Look at Memorization in Deep Networks
• Image Classification with Deep Learning in the Presence of Noisy Labels: A Survey
• WebVision Database: Visual Learning and Understanding from Web Data
• Dimensionality-Driven Learning with Noisy Labels
• Towards Understanding the Role of Over-Parametrization in Generalization of Neural Networks
• How does Disagreement Help Generalization against Label Corruption?
• Gradient Descent with Early Stopping is Provably Robust to Label Noise for Overparameterized Neural Networks
• Learning with Bad Training Data via Iterative Trimmed Loss Minimization
• Predicting the Generalization Gap in Deep Networks with Margin Distributions
• Fisher-Rao Metric, Geometry, and Complexity of Neural Networks
• Understanding and Utilizing Deep Neural Networks Trained with Noisy Labels
• L_DMI: An Information-theoretic Noise-robust Loss Function
• Searching to Exploit Memorization Effect in Learning from Corrupted Labels

Cited by in corpus (17)

• Contrast to Divide: Self-Supervised Pre-Training for Learning with Noisy Labels
• Training Data Influence Analysis and Estimation: A Survey
• FINE Samples for Learning with Noisy Labels
• Data Heterogeneity Modeling for Trustworthy Machine Learning
• DyGen: Learning from Noisy Labels via Dynamics-Enhanced Generative Modeling
• Faster Meta Update Strategy for Noise-Robust Deep Learning
• Confidence Adaptive Regularization for Deep Learning with Noisy Labels
• Instance Correction for Learning with Open-set Noisy Labels
• Deep Learning with Label Differential Privacy
• Evolution-aware VAriance (EVA) Coreset Selection for Medical Image Classification
• Learning with Noisy Labels by Efficient Transition Matrix Estimation to Combat Label Miscorrection
• Calibrated Top-1 Uncertainty estimates for classification by score based models
• Self-semi-supervised Learning to Learn from NoisyLabeled Data
• A Tale Of Two Long Tails
• Training Dynamic based data filtering may not work for NLP datasets
• Improving Generalization of Deep Fault Detection Models in the Presence of Mislabeled Data
• Tips and Tricks for Webly-Supervised Fine-Grained Recognition: Learning from the WebFG 2020 Challenge