Recent studies show that pretraining a deep neural network with fine-grained labeled data, followed by fine-tuning on coarse-labeled data for downstream tasks, often yields better generalization than pretraining with coarse-labeled data. While there is ample empirical evidence supporting this, the theoretical justification remains an open problem. This paper addresses this gap by introducing a "hierarchical multi-view" structure to confine the input data distribution. Under this framework, we prove that: 1) coarse-grained pretraining only allows a neural network to learn the common features well, while 2) fine-grained pretraining helps the network learn the rare features in addition to the common ones, leading to improved accuracy on hard downstream test samples.
@article{arxiv.2410.23129,
title = {Why Fine-grained Labels in Pretraining Benefit Generalization?},
author = {Guan Zhe Hong and Yin Cui and Ariel Fuxman and Stanley Chan and Enming Luo},
journal= {arXiv preprint arXiv:2410.23129},
year = {2024}
}
Comments
arXiv admin note: substantial text overlap with arXiv:2303.16887