Training BatchNorm and Only BatchNorm: On the Expressive Power of Random Features in CNNsDownload PDF

Published: 12 Jan 2021, Last Modified: 22 Oct 2023ICLR 2021 PosterReaders: Everyone
Keywords: affine parameters, random features, batchnorm
Abstract: A wide variety of deep learning techniques from style transfer to multitask learning rely on training affine transformations of features. Most prominent among these is the popular feature normalization technique BatchNorm, which normalizes activations and then subsequently applies a learned affine transform. In this paper, we aim to understand the role and expressive power of affine parameters used to transform features in this way. To isolate the contribution of these parameters from that of the learned features they transform, we investigate the performance achieved when training only these parameters in BatchNorm and freezing all weights at their random initializations. Doing so leads to surprisingly high performance considering the significant limitations that this style of training imposes. For example, sufficiently deep ResNets reach 82% (CIFAR-10) and 32% (ImageNet, top-5) accuracy in this configuration, far higher than when training an equivalent number of randomly chosen parameters elsewhere in the network. BatchNorm achieves this performance in part by naturally learning to disable around a third of the random features. Not only do these results highlight the expressive power of affine parameters in deep learning, but - in a broader sense - they characterize the expressive power of neural networks constructed simply by shifting and rescaling random features.
One-sentence Summary: We study the role and expressive power of learned affine parameters that transform features by freezing all weights at their random initializations and training only BatchNorm.
Code Of Ethics: I acknowledge that I and all co-authors of this work have read and commit to adhering to the ICLR Code of Ethics
Code: [![Papers with Code](/images/pwc_icon.svg) 4 community implementations](https://paperswithcode.com/paper/?openreview=vYeQQ29Tbvx)
Community Implementations: [![CatalyzeX](/images/catalyzex_icon.svg) 4 code implementations](https://www.catalyzex.com/paper/arxiv:2003.00152/code)
12 Replies

Loading