Deep learning algorithms can fare poorly when the training dataset suffers from heavy class-imbalance but the testing criterion requires good generalization on less frequent classes. We design two novel methods to improve performance in such scenarios. First, we propose a theoretically-principled label-distribution-aware margin (LDAM) loss motivated by minimizing a margin-based generalization bound. This loss replaces the standard cross-entropy objective during training and can be applied with prior strategies for training with class-imbalance such as re-weighting or re-sampling. Second, we propose a simple, yet effective, training schedule that defers re-weighting until after the initial stage, allowing the model to learn an initial representation while avoiding some of the complications associated with re-weighting or re-sampling. We test our methods on several benchmark vision tasks including the real-world imbalanced dataset iNaturalist 2018. Our experiments show that either of these methods alone can already improve over existing techniques and their combination achieves even better performance gains.

LDAM aims to put regularization on the margins (i.e. the minimum distance of data samples to the decision boundary) of minority classes in order to improve the generalizability of the model towards minority classes during the test time, in which the value of margin is set to be proportional to the number of samples for each class thus the LDAM is label-distribution-aware. DRW runs reweighting and LDAM with smaller learning rate in order to perform fine-tuning on the model after an initial stage of training. Although without any theoretical justification, its efficacy is successfully proven across various experiments. These ideas are novel and shown to provide better superior performance, even avoid overfitting for frequent classes, in comparison to naive re-weighting/re-sampling techniques and other baselines (e.g. It is also required to have an analysis on the sensitivity of performance with respect to C. Additionally, as in both stages of Algorithm.1 LDAM is used, should there be different values of C? - How is the LDAM-HG-DRS in Table.1 implemented?

The authors propose to solve imbalanced classification problems by requiring uneven margins for each class of examples. The reviewers agree that the proposed view is novel and comes with some insights from the theoretical side. The authors then design a new loss to achieve the uneven margins and apply the loss within a two-stage algorithm to achieve promising performance on some image data sets. The promising performance on the important problem of imbalanced classification makes the paper sufficiently interesting for the NeurIPS audience. The authors are somehow encouraged to clarify the gap between the theoretical results on binary classification and the algorithmic results on multi-class classification.

Deep learning algorithms can fare poorly when the training dataset suffers from heavy class-imbalance but the testing criterion requires good generalization on less frequent classes. We design two novel methods to improve performance in such scenarios. First, we propose a theoretically-principled label-distribution-aware margin (LDAM) loss motivated by minimizing a margin-based generalization bound. This loss replaces the standard cross-entropy objective during training and can be applied with prior strategies for training with class-imbalance such as re-weighting or re-sampling. Second, we propose a simple, yet effective, training schedule that defers re-weighting until after the initial stage, allowing the model to learn an initial representation while avoiding some of the complications associated with re-weighting or re-sampling.

A learning algorithm referred to as Maximum Margin (MM) is proposed for considering the class-imbalance data learning issue: the trained model tends to predict the majority of classes rather than the minority ones. That is, underfitting for minority classes seems to be one of the challenges of generalization. For a good generalization of the minority classes, we design a new Maximum Margin (MM) loss function, motivated by minimizing a margin-based generalization bound through the shifting decision bound. The theoretically-principled label-distribution-aware margin (LDAM) loss was successfully applied with prior strategies such as re-weighting or re-sampling along with the effective training schedule. However, they did not investigate the maximum margin loss function yet. In this study, we investigate the performances of two types of hard maximum margin-based decision boundary shift with LDAM's training schedule on artificially imbalanced CIFAR-10/100 for fair comparisons and effectiveness.

Deep learning algorithms can fare poorly when the training dataset suffers from heavy class-imbalance but the testing criterion requires good generalization on less frequent classes. We design two novel methods to improve performance in such scenarios. First, we propose a theoretically-principled label-distribution-aware margin (LDAM) loss motivated by minimizing a margin-based generalization bound. This loss replaces the standard cross-entropy objective during training and can be applied with prior strategies for training with class-imbalance such as re-weighting or re-sampling. Second, we propose a simple, yet effective, training schedule that defers re-weighting until after the initial stage, allowing the model to learn an initial representation while avoiding some of the complications associated with re-weighting or re-sampling.