Micro-Batch Training with Batch-Channel Normalization and Weight Standardization

TL;DR

This paper introduces Weight Standardization and Batch-Channel Normalization to enhance micro-batch training in deep networks, enabling models trained with very small batch sizes to achieve performance comparable to large-batch training.

Contribution

The paper proposes novel normalization techniques, WS and BCN, specifically designed to improve the effectiveness of micro-batch training in various computer vision tasks.

Findings

WS and BCN significantly improve micro-batch training performance.

Models with WS and BCN can match or outperform large-batch BN training.

Validated across multiple vision tasks with consistent results.

Abstract

Batch Normalization (BN) has become an out-of-box technique to improve deep network training. However, its effectiveness is limited for micro-batch training, i.e., each GPU typically has only 1-2 images for training, which is inevitable for many computer vision tasks, e.g., object detection and semantic segmentation, constrained by memory consumption. To address this issue, we propose Weight Standardization (WS) and Batch-Channel Normalization (BCN) to bring two success factors of BN into micro-batch training: 1) the smoothing effects on the loss landscape and 2) the ability to avoid harmful elimination singularities along the training trajectory. WS standardizes the weights in convolutional layers to smooth the loss landscape by reducing the Lipschitz constants of the loss and the gradients; BCN combines batch and channel normalizations and leverages estimated statistics of the activations in convolutional layers to keep networks away from elimination singularities. We validate WS and BCN on comprehensive computer vision tasks, including image classification, object detection, instance segmentation, video recognition and semantic segmentation. All experimental results consistently show that WS and BCN improve micro-batch training significantly. Moreover, using WS and BCN with micro-batch training is even able to match or outperform the performances of BN with large-batch training.