Improved Regularization of Convolutional Neural Networks with Cutout

2017 article devries2017cutout ⚠ Needs review - 1 field differ

Fields with differences: venue. Compare local vs external BibTeX below.

Authors

Terrance DeVries, Graham W. Taylor

Venue

arXiv preprint

Abstract

Introduces Cutout, a regularization technique that randomly masks square regions of input images during training. Inspired by dropout but applied to inputs, encouraging models to learn from partially visible objects.

BibTeX

Local Entry

@article{devries2017cutout,
  title = {Improved Regularization of Convolutional Neural Networks with Cutout},
  author = {Terrance DeVries and Graham W. Taylor},
  year = {2017},
  journal = {arXiv preprint},
  url = {https://arxiv.org/abs/1708.04552},
  abstract = {Introduces Cutout, a regularization technique that randomly masks square regions of input images during training. Inspired by dropout but applied to inputs, encouraging models to learn from partially visible objects.}
}

From AUTO:OPENALEX

@article{devries2017cutout,
  title = {Improved Regularization of Convolutional Neural Networks with Cutout},
  author = {Terrance DeVries and Graham W. Taylor},
  year = {2017},
  journal = {arXiv (Cornell University)},
  doi = {10.48550/arxiv.1708.04552}
}