Fetching the paper…

Learning Adversarially Robust Representations via Worst-Case Mutual Information Maximization · Around