Uncertainty Estimation via Stochastic Batch Normalization
Abstract
In this work, we investigate Batch Normalization technique and propose its probabilistic interpretation. We propose a probabilistic model and show that Batch Normalization maximazes the lower bound of its marginalized log-likelihood. Then, according to the new probabilistic model, we design an algorithm which acts consistently during train and test. However, inference becomes computationally inefficient. To reduce memory and computational cost, we propose Stochastic Batch Normalization -- an efficient approximation of proper inference procedure. This method provides us with a scalable uncertainty estimation technique. We demonstrate the performance of Stochastic Batch Normalization on popular architectures (including deep convolutional architectures: VGG-like and ResNets) for MNIST and CIFAR-10 datasets.
Cite
@article{arxiv.1802.04893,
title = {Uncertainty Estimation via Stochastic Batch Normalization},
author = {Andrei Atanov and Arsenii Ashukha and Dmitry Molchanov and Kirill Neklyudov and Dmitry Vetrov},
journal= {arXiv preprint arXiv:1802.04893},
year = {2018}
}
Comments
Under review as a workshop paper at ICLR 2018