Filter Response Normalization Layer: Eliminating Batch Dependence in the Training of Deep Neural Networks

TL;DR

The paper introduces the Filter Response Normalization (FRN) layer, a new normalization technique that eliminates batch dependence, improving training stability and accuracy across various batch sizes in deep neural networks.

Contribution

The FRN layer combines normalization and activation, operating independently on each activation channel, outperforming Batch Normalization and alternatives across different batch sizes.

Findings

FRN outperforms BN by 0.7-1.0% in ImageNet classification with large batches.

FRN exceeds Group Normalization by over 1% in small batch regimes.

FRN improves object detection accuracy on COCO by 0.3-0.5% across batch sizes.

Abstract

Batch Normalization (BN) uses mini-batch statistics to normalize the activations during training, introducing dependence between mini-batch elements. This dependency can hurt the performance if the mini-batch size is too small, or if the elements are correlated. Several alternatives, such as Batch Renormalization and Group Normalization (GN), have been proposed to address this issue. However, they either do not match the performance of BN for large batches, or still exhibit degradation in performance for smaller batches, or introduce artificial constraints on the model architecture. In this paper we propose the Filter Response Normalization (FRN) layer, a novel combination of a normalization and an activation function, that can be used as a replacement for other normalizations and activations. Our method operates on each activation channel of each batch element independently, eliminating the dependency on other batch elements. Our method outperforms BN and other alternatives in a variety of settings for all batch sizes. FRN layer performs $\approx 0.7-1.0\%$ better than BN on top-1 validation accuracy with large mini-batch sizes for Imagenet classification using InceptionV3 and ResnetV2-50 architectures. Further, it performs $>1\%$ better than GN on the same problem in the small mini-batch size regime. For object detection problem on COCO dataset, FRN layer outperforms all other methods by at least $0.3-0.5\%$ in all batch size regimes.