Paper ID: 2112.12299

A Robust Initialization of Residual Blocks for Effective ResNet Training without Batch Normalization

Enrico Civitelli, Alessio Sortino, Matteo Lapucci, Francesco Bagattini, Giulio Galvan

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.

Submitted: Dec 23, 2021