A Robust Initialization of Residual Blocks for Effective ResNet Training without Batch Normalization
Abstract
Batch Normalization is an essential component of all state-of-the-art neural networks architectures. However, since it introduces many practical issues, much recent research has been devoted to designing normalization-free architectures. In this paper, we show that weights initialization is key to train ResNet-like normalization-free networks. In particular, we propose a slight modification to the summation operation of a block output to the skip-connection branch, so that the whole network is correctly initialized. We show that this modified architecture achieves competitive results on CIFAR-10, CIFAR-100 and ImageNet without further regularization nor algorithmic modifications.
Cite
@article{arxiv.2112.12299,
title = {A Robust Initialization of Residual Blocks for Effective ResNet Training without Batch Normalization},
author = {Enrico Civitelli and Alessio Sortino and Matteo Lapucci and Francesco Bagattini and Giulio Galvan},
journal= {arXiv preprint arXiv:2112.12299},
year = {2023}
}
Comments
16 pages (4 pages of supplementary material), 9 figures, 2 table