Leveraging Angular Information Between Feature and Classifier for Long-tailed Learning: A Prediction Reformulation Approach
This addresses the imbalance issue in long-tailed learning for computer vision applications, offering a novel approach that improves performance without re-balancing classifier weights.
The paper tackles the problem of deep neural networks struggling with long-tailed image datasets by reformulating recognition probabilities using angles between data features and classifier weights, achieving state-of-the-art performance on CIFAR10/100-LT and ImageNet-LT without pretraining.
Deep neural networks still struggle on long-tailed image datasets, and one of the reasons is that the imbalance of training data across categories leads to the imbalance of trained model parameters. Motivated by the empirical findings that trained classifiers yield larger weight norms in head classes, we propose to reformulate the recognition probabilities through included angles without re-balancing the classifier weights. Specifically, we calculate the angles between the data feature and the class-wise classifier weights to obtain angle-based prediction results. Inspired by the performance improvement of the predictive form reformulation and the outstanding performance of the widely used two-stage learning framework, we explore the different properties of this angular prediction and propose novel modules to improve the performance of different components in the framework. Our method is able to obtain the best performance among peer methods without pretraining on CIFAR10/100-LT and ImageNet-LT. Source code will be made publicly available.