Rethinking Normalization and Elimination Singularity in Neural Networks
arXiv:1911.09738
Abstract
In this paper, we study normalization methods for neural networks from the perspective of elimination singularity. Elimination singularities correspond to the points on the training trajectory where neurons become consistently deactivated. They cause degenerate manifolds in the loss landscape which will slow down training and harm model performances. We show that channel-based normalizations (e.g. Layer Normalization and Group Normalization) are unable to guarantee a far distance from elimination singularities, in contrast with Batch Normalization which by design avoids models from getting too close to them. To address this issue, we propose BatchChannel Normalization (BCN), which uses batch knowledge to avoid the elimination singularities in the training of channel-normalized models. Unlike Batch Normalization, BCN is able to run in both large-batch and micro-batch training settings. The effectiveness of BCN is verified on many tasks, including image classification, object detection, instance segmentation, and semantic segmentation. The code is here: https://github.com/joe-siyuan-qiao/Batch-Channel-Normalization.
arXiv admin note: substantial text overlap with arXiv:1903.10520
References in corpus (16)
- Batch Normalization: Accelerating Deep Network Training by Reducing Internal Covariate Shift
- Faster R-CNN: Towards Real-Time Object Detection with Region Proposal Networks
- Rethinking Atrous Convolution for Semantic Image Segmentation
- Instance Normalization: The Missing Ingredient for Fast Stylization
- MMDetection: Open MMLab Detection Toolbox and Benchmark
- Batch Renormalization: Towards Reducing Minibatch Dependence in Batch-Normalized Models
- On the importance of single directions for generalization
- Differentiable Learning-to-Normalize via Switchable Normalization
- Micro-Batch Training with Batch-Channel Normalization and Weight Standardization
- Skip Connections Eliminate Singularities
- Knowledge Distillation in Generations: More Tolerant Teachers Educate Better Students
- Theoretical Analysis of Auto Rate-Tuning by Batch Normalization
- A Mean Field Theory of Batch Normalization
- Towards Understanding Regularization in Batch Normalization
- Exponential convergence rates for Batch Normalization: The power of length-direction decoupling in non-convex optimization
- SSN: Learning Sparse Switchable Normalization via SparsestMax