Distilling Virtual Examples for Long-tailed Recognition
This work addresses the problem of imbalanced class distributions in visual recognition for machine learning practitioners, offering an incremental improvement over existing methods.
The paper tackles long-tailed visual recognition by proposing a knowledge distillation method called DiVE, which treats teacher model predictions as virtual examples and flattens their distribution to improve tail class performance, achieving significant gains over state-of-the-art methods on benchmark datasets like iNaturalist.
We tackle the long-tailed visual recognition problem from the knowledge distillation perspective by proposing a Distill the Virtual Examples (DiVE) method. Specifically, by treating the predictions of a teacher model as virtual examples, we prove that distilling from these virtual examples is equivalent to label distribution learning under certain constraints. We show that when the virtual example distribution becomes flatter than the original input distribution, the under-represented tail classes will receive significant improvements, which is crucial in long-tailed recognition. The proposed DiVE method can explicitly tune the virtual example distribution to become flat. Extensive experiments on three benchmark datasets, including the large-scale iNaturalist ones, justify that the proposed DiVE method can significantly outperform state-of-the-art methods. Furthermore, additional analyses and experiments verify the virtual example interpretation, and demonstrate the effectiveness of tailored designs in DiVE for long-tailed problems.