Simplifying Neural Network Training Under Class Imbalance
Abstract
Real-world datasets are often highly class-imbalanced, which can adversely impact the performance of deep learning models. The majority of research on training neural networks under class imbalance has focused on specialized loss functions and sampling techniques. Notably, we demonstrate that simply tuning existing components of standard deep learning pipelines, such as the batch size, data augmentation, architecture size, pre-training, optimizer, and label smoothing, can achieve state-of-the-art performance without any specialized loss functions or samplers. We also provide key prescriptions and considerations for training under class imbalance, and an understanding of why imbalance methods succeed or fail.
Cite
Text
Shwartz-Ziv et al. "Simplifying Neural Network Training Under Class Imbalance." Neural Information Processing Systems, 2023.Markdown
[Shwartz-Ziv et al. "Simplifying Neural Network Training Under Class Imbalance." Neural Information Processing Systems, 2023.](https://mlanthology.org/neurips/2023/shwartzziv2023neurips-simplifying/)BibTeX
@inproceedings{shwartzziv2023neurips-simplifying,
title = {{Simplifying Neural Network Training Under Class Imbalance}},
author = {Shwartz-Ziv, Ravid and Goldblum, Micah and Li, Yucen and Bruss, C. Bayan and Wilson, Andrew G},
booktitle = {Neural Information Processing Systems},
year = {2023},
url = {https://mlanthology.org/neurips/2023/shwartzziv2023neurips-simplifying/}
}