NeurIPS 2019
Sun Dec 8th through Sat the 14th, 2019 at Vancouver Convention Center
Paper ID:3462
Title:The Normalization Method for Alleviating Pathological Sharpness in Wide Neural Networks

The paper is well-written paper and analyzes how signals propagate in random neural networks. It does so by analyzing mean and variance of activations and gradients, given random inputs and weights. The technical contributions are okay, and the analysis leads to new insights on the use of batch/layer normalization.