Transfer Learning with Fewer ImageNet Classes

Abstract

Though much previous work tried to uncover the best practices for transfer learning, much is left unexplored. Our preliminary work explores the effect of removing a portion of the ImageNet classes with low per-class validation accuracy on the accuracy of the remaining classes. Furthermore, we explore if models trained with a reduced number of classes are suitable for transfer learning.

PDF NeurIPSW OpenReview Semantic Scholar

Cite

Text

Kucer and Oyen. "Transfer Learning with Fewer ImageNet Classes." NeurIPS 2021 Workshops: ImageNet_PPF, 2021.

Markdown

[Kucer and Oyen. "Transfer Learning with Fewer ImageNet Classes." NeurIPS 2021 Workshops: ImageNet_PPF, 2021.](https://mlanthology.org/neuripsw/2021/kucer2021neuripsw-transfer/)

BibTeX

@inproceedings{kucer2021neuripsw-transfer,
  title     = {{Transfer Learning with Fewer ImageNet Classes}},
  author    = {Kucer, Michal and Oyen, Diane},
  booktitle = {NeurIPS 2021 Workshops: ImageNet_PPF},
  year      = {2021},
  url       = {https://mlanthology.org/neuripsw/2021/kucer2021neuripsw-transfer/}
}