CEB Improves Model Robustness

Ian Fischer, Alex A. Alemi

Keywords: Information Theory, Adversarial Robustness

TL;DR: Training your model with the Conditional Entropy Bottleneck is easy and makes your model more robust.

Abstract: We demonstrate that the Conditional Entropy Bottleneck (CEB) can improve model robustness. CEB is an easy strategy to implement and works in tandem with data augmentation procedures. We report results of a large scale adversarial robustness study on CIFAR-10, as well as the IMAGENET-C Common Corruptions Benchmark.

Community Implementations: [ 1 code implementation](https://www.catalyzex.com/paper/arxiv:2002.05380/code)