How Does Batch Normalization Help Optimization?
This addresses a fundamental question in machine learning for researchers and practitioners, providing new insights into a widely used technique.
The paper tackled the problem of understanding why Batch Normalization (BatchNorm) improves deep neural network training, and found that its success is due to smoothing the optimization landscape rather than reducing internal covariate shift, leading to faster and more stable training.
Batch Normalization (BatchNorm) is a widely adopted technique that enables faster and more stable training of deep neural networks (DNNs). Despite its pervasiveness, the exact reasons for BatchNorm's effectiveness are still poorly understood. The popular belief is that this effectiveness stems from controlling the change of the layers' input distributions during training to reduce the so-called "internal covariate shift". In this work, we demonstrate that such distributional stability of layer inputs has little to do with the success of BatchNorm. Instead, we uncover a more fundamental impact of BatchNorm on the training process: it makes the optimization landscape significantly smoother. This smoothness induces a more predictive and stable behavior of the gradients, allowing for faster training.