Mind the Pool: Convolutional Neural Networks Can Overfit Input Size
Bilal Alsallakh, David Yan, Narine Kokhlikyan, Vivek Miglani, Orion Reblitz-Richardson, Pamela Bhattacharya
Abstract
We demonstrate how convolutional neural networks can overfit the input size: The accuracy drops significantly when using certain sizes, compared with favorable ones. This issue is inherent to pooling arithmetic, with standard downsampling layers playing a major role in favoring certain input sizes and skewing the weights accordingly. We present a solution to this problem by depriving these layers from the arithmetic cues they use to overfit the input size. Through various examples, we show how our proposed spatially-balanced pooling improves the generalization of the network to arbitrary input sizes and its robustness to translational shifts.
BibTeX
@inproceedings{
alsallakh2023mind,
title={Mind the Pool: Convolutional Neural Networks Can Overfit Input Size},
author={Bilal Alsallakh and David Yan and Narine Kokhlikyan and Vivek Miglani and Orion Reblitz-Richardson and Pamela Bhattacharya},
booktitle={The Eleventh International Conference on Learning Representations },
year={2023},
url={https://openreview.net/forum?id=cWmtUcsYC3V}
}