Feature-Balanced Loss for Long-Tailed Visual Recognition
This work addresses the long-tailed recognition problem in computer vision, which is a common issue in real-world datasets, and is incremental as it builds on existing methods by focusing on feature space adjustments.
The paper tackles the problem of performance degradation in deep neural networks trained on long-tailed data by proposing a feature-balanced loss that encourages larger feature norms for tail classes using curriculum learning, achieving superior performance gains on multiple benchmarks.
Deep neural networks frequently suffer from performance degradation when the training data is long-tailed because several majority classes dominate the training, resulting in a biased model. Recent studies have made a great effort in solving this issue by obtaining good representations from data space, but few of them pay attention to the influence of feature norm on the predicted results. In this paper, we therefore address the long-tailed problem from feature space and thereby propose the feature-balanced loss. Specifically, we encourage larger feature norms of tail classes by giving them relatively stronger stimuli. Moreover, the stimuli intensity is gradually increased in the way of curriculum learning, which improves the generalization of the tail classes, meanwhile maintaining the performance of the head classes. Extensive experiments on multiple popular long-tailed recognition benchmarks demonstrate that the feature-balanced loss achieves superior performance gains compared with the state-of-the-art methods.