Skip to yearly menu bar Skip to main content


Differentiable Dynamic Normalization for Learning Deep Representation

Ping Luo · Peng Zhanglin · Shao Wenqi · Zhang ruimao · Ren jiamin · Wu lingyun

Pacific Ballroom #25

Keywords: [ Computer Vision ] [ Algorithms ]


This work presents Dynamic Normalization (DN), which is able to learn arbitrary normalization operations for different convolutional layers in a deep ConvNet. Unlike existing normalization approaches that predefined computations of the statistics (mean and variance), DN learns to estimate them. DN has several appealing benefits. First, it adapts to various networks, tasks, and batch sizes. Second, it can be easily implemented and trained in a differentiable end-to-end manner with merely small number of parameters. Third, its matrix formulation represents a wide range of normalization methods, shedding light on analyzing them theoretically. Extensive studies show that DN outperforms its counterparts in CIFAR10 and ImageNet.

Live content is unavailable. Log in and register to view live content