Open-source software:

DRIBO: Robust Deep Reinforcement Learning via Multi-View Information Bottleneck

  • A new multi-view information bottleneck (MIB) objective that maximizes the mutual information between sequences of observations and sequences of representations while reducing the task-irrelevant information identified through the multi-view observations.
  • State-of-the-art results on the DeepMind Control Suite (with videos playing in the background as visual distractors) and the Procgen Benchmark (in terms of the performance of generalizing the learned policy to unseen levels).
  • Github link:
  • Related publication:
    1. Jiameng Fan and Wenchao Li. DRIBO: Robust Deep Reinforcement Learning via Multi-View Information Bottleneck. The 39th International Conference on Machine Learning (ICML), 2022 (to appear).

AdvIBP: Certified Adversarial Training by Combining Adversarial Training and Provable Robustness Verification in a Principled Way

  • State-of-the-art verified (certified) error on MNIST and CIFAR: for MNIST, 6.60% at epsilon=0.3 and 12.30% at epsilon=0.4 (for L_infinity norm perturbations); and for CIFAR, 66.57% at epsilon=8/255 and 76.05% at epsilon=16/255 (also for L_infinity norm perturbations).
  • Github link:
  • Related publication(s):
    1. Jiameng Fan and Wenchao Li. Adversarial Training and Provable Robustness: A Tale of Two Objectives. The 35th AAAI Conference on Artificial Intelligence (AAAI), February 2021. 

ReachNN*: Reachability Analysis Tool of Neural Network Controlled Systems (NNCSs)