Solving The Long-Tailed Problem via Intra- and Inter-Category Balance
This work addresses performance degradation in real-world datasets with long-tailed distributions for visual recognition applications, representing an incremental improvement over existing re-sampling and re-weighting methods.
The paper tackles the long-tailed distribution problem in visual recognition by proposing a gradient harmonized mechanism with intra- and inter-category balance strategies to address hard examples in head classes and decision boundary shifts, resulting in consistent outperformance over other approaches on all datasets.
Benchmark datasets for visual recognition assume that data is uniformly distributed, while real-world datasets obey long-tailed distribution. Current approaches handle the long-tailed problem to transform the long-tailed dataset to uniform distribution by re-sampling or re-weighting strategies. These approaches emphasize the tail classes but ignore the hard examples in head classes, which result in performance degradation. In this paper, we propose a novel gradient harmonized mechanism with category-wise adaptive precision to decouple the difficulty and sample size imbalance in the long-tailed problem, which are correspondingly solved via intra- and inter-category balance strategies. Specifically, intra-category balance focuses on the hard examples in each category to optimize the decision boundary, while inter-category balance aims to correct the shift of decision boundary by taking each category as a unit. Extensive experiments demonstrate that the proposed method consistently outperforms other approaches on all the datasets.