arXiv 2102.06171
High-Performance Large-Scale Image Recognition Without Normalization
By Andrew Brock, Soham De, et al.
Published 2021-02-11
Citation lineage
Review the prior work and downstream research connected to this paper.
Batch normalization is a key component of most image classification models, but it has many undesirable properties stemming from its dependence on the batch size and interactions between examples. Although recent work has succeeded in training deep ResNets without normalization layers, these models do not match the test accuracies of the best batch-normalized networks, and are often unstable for large learning rates…