CSA

This is the source code for the paper: How Re-sampling Helps for Long-Tail Learning?

For experiments on MNIST-LT, Fashion-LT and CMNIST-LT, please refer to CMNIST-LT.

For implementation of Context-Shift Augmentation (CSA), please refer to CSA.

References:

Jiang-Xin Shi, Tong Wei, Yuke Xiang, Yu-Feng Li. How Re-sampling Helps for Long-Tail Learning?. In: Advances in Neural Information Processing Systems (NeurIPS'23).