Dataset balancing can hurt model performance

Abstract

Machine learning from training data with a skewed distribution of examples per class can lead to models that favor performance on common classes at the expense of performance on rare ones. AudioSet has a very wide range of priors over its 527 sound event classes. Classification performance on AudioSet is usually evaluated by a simple average over per-class metrics, meaning that performance on rare classes is equal in importance to the performance on common ones. Several recent papers have used dataset balancing techniques to improve performance on AudioSet. We find, however, that while balancing improves performance on the public AudioSet evaluation data it simultaneously hurts performance on an unpublished evaluation set collected under the same conditions. By varying the degree of balancing, we show that its benefits are fragile and depend on the evaluation set. We also do not find evidence indicating that balancing improves rare class performance relative to common classes. We therefore caution against blind application of balancing, as well as against paying too much attention to small improvements on a public evaluation set.

arXiv comment5 pages, 3 figures, ICASSP 2023

References in corpus (6)

• Adam: A Method for Stochastic Optimization
• An Image is Worth 16x16 Words: Transformers for Image Recognition at Scale
• Distilling the Knowledge in a Neural Network
• mixup: Beyond Empirical Risk Minimization
• SpecAugment: A Simple Data Augmentation Method for Automatic Speech Recognition
• Solution for Large-Scale Hierarchical Object Detection Datasets with Incomplete Annotation and Data Imbalance