Difficulty-aware Balancing Margin Loss for Long-tailed Recognition
This work addresses the challenge of accurate recognition for rare classes in imbalanced datasets, which is incremental as it builds on existing rebalancing methods by incorporating instance-level difficulty.
The paper tackles the problem of deep neural networks struggling with class imbalance in long-tailed recognition by proposing a difficulty-aware balancing margin loss that considers both class imbalance and instance difficulty, resulting in consistent performance improvements across benchmarks.
When trained with severely imbalanced data, deep neural networks often struggle to accurately recognize classes with only a few samples. Previous studies in long-tailed recognition have attempted to rebalance biased learning using known sample distributions, primarily addressing different classification difficulties at the class level. However, these approaches often overlook the instance difficulty variation within each class. In this paper, we propose a difficulty-aware balancing margin (DBM) loss, which considers both class imbalance and instance difficulty. DBM loss comprises two components: a class-wise margin to mitigate learning bias caused by imbalanced class frequencies, and an instance-wise margin assigned to hard positive samples based on their individual difficulty. DBM loss improves class discriminativity by assigning larger margins to more difficult samples. Our method seamlessly combines with existing approaches and consistently improves performance across various long-tailed recognition benchmarks.