Class-Balanced Loss Based on Effective Number of Samples

Code

• github.com/richardaecn/class-balanced-loss
  OfficialIn papertf★ 0
• github.com/frgfm/Holocron
  pytorch★ 328
• github.com/bazinga699/ncl
  pytorch★ 86
• github.com/lijian16/fcc
  pytorch★ 23
• github.com/feidfoe/AdjustBnd4Imbalance
  pytorch★ 19
• github.com/statsu1990/yoto_class_balanced_loss
  pytorch★ 0
• github.com/MindSpore-scientific/code-3/tree/main/Class-balanced-loss-pytorch-master
  mindspore★ 0
• github.com/tiagoCuervo/JapaNet
  tf★ 0
• github.com/vandit15/Class-balanced-loss-pytorch
  pytorch★ 0
• github.com/MindCode-4/code-11/tree/main/Class-balanced-loss-pytorch-master
  mindspore★ 0

Abstract

With the rapid increase of large-scale, real-world datasets, it becomes critical to address the problem of long-tailed data distribution (i.e., a few classes account for most of the data, while most classes are under-represented). Existing solutions typically adopt class re-balancing strategies such as re-sampling and re-weighting based on the number of observations for each class. In this work, we argue that as the number of samples increases, the additional benefit of a newly added data point will diminish. We introduce a novel theoretical framework to measure data overlap by associating with each sample a small neighboring region rather than a single point. The effective number of samples is defined as the volume of samples and can be calculated by a simple formula (1-^n)/(1-), where n is the number of samples and [0,1) is a hyperparameter. We design a re-weighting scheme that uses the effective number of samples for each class to re-balance the loss, thereby yielding a class-balanced loss. Comprehensive experiments are conducted on artificially induced long-tailed CIFAR datasets and large-scale datasets including ImageNet and iNaturalist. Our results show that when trained with the proposed class-balanced loss, the network is able to achieve significant performance gains on long-tailed datasets.

Tasks

Benchmark Results

Reproductions