# Regret-Based Defense in Adversarial Reinforcement Learning

arXiv: https://arxiv.org/abs/2302.06912 · HTML: https://mlsanity.com/paper/2302.06912/

## Key facts

- Cited by 3 papers in the map (data through 14 Oct 2026).
- Builds on 22 of its 32 references that are in the map.
- First posted on arXiv 14 Feb 2023.

## Builds on (most cited first, 22 of 22)

- [Explaining and Harnessing Adversarial Examples](https://mlsanity.com/paper/1412.6572/) (2014)
- [Towards Deep Learning Models Resistant to Adversarial Attacks](https://mlsanity.com/paper/1706.06083/) (2017)
- [Playing Atari with Deep Reinforcement Learning](https://mlsanity.com/paper/1312.5602/) (2013)
- [Domain-Adversarial Training of Neural Networks](https://mlsanity.com/paper/1505.07818/) (2015)
- [Fast is better than free: Revisiting adversarial training](https://mlsanity.com/paper/2001.03994/) (2020)
- [Adversarial Training for Free!](https://mlsanity.com/paper/1904.12843/) (2019)
- [Robust Adversarial Reinforcement Learning](https://mlsanity.com/paper/1703.02702/) (2017)
- [Adversarial Attacks on Neural Network Policies](https://mlsanity.com/paper/1702.02284/) (2017)
- [Recent Advances in Adversarial Training for Adversarial Robustness](https://mlsanity.com/paper/2102.01356/) (2021)
- [Robust Deep Reinforcement Learning against Adversarial Perturbations on State Observations](https://mlsanity.com/paper/2003.08938/) (2020)
- [Adversarial Policies: Attacking Deep Reinforcement Learning](https://mlsanity.com/paper/1905.10615/) (2019)
- [Understanding and Improving Fast Adversarial Training](https://mlsanity.com/paper/2007.02617/) (2020)
- [Robust Deep Reinforcement Learning with Adversarial Attacks](https://mlsanity.com/paper/1712.03632/) (2017)
- [EPOpt: Learning Robust Neural Network Policies Using Model Ensembles](https://mlsanity.com/paper/1610.01283/) (2016)
- [Universal Adversarial Training](https://mlsanity.com/paper/1811.11304/) (2018)
- [Robust Deep Reinforcement Learning through Adversarial Loss](https://mlsanity.com/paper/2008.01976/) (2020)
- [Stealthy and Efficient Adversarial Attacks against Deep Reinforcement Learning](https://mlsanity.com/paper/2005.07099/) (2020)
- [Who Is the Strongest Enemy? Towards Optimal and Efficient Evasion Attacks in Deep RL](https://mlsanity.com/paper/2106.05087/) (2021)
- [Certified Adversarial Robustness for Deep Reinforcement Learning](https://mlsanity.com/paper/1910.12908/) (2019)
- [Efficient Adversarial Training without Attacking: Worst-Case-Aware Robust Reinforcement Learning](https://mlsanity.com/paper/2210.05927/) (2022)
- [Robustifying Reinforcement Learning Agents via Action Space Adversarial Training](https://mlsanity.com/paper/2007.07176/) (2020)
- [Transfer of Adversarial Robustness Between Perturbation Types](https://mlsanity.com/paper/1905.01034/) (2019)

## Built on by (most cited first, 3 of 3)

- [Robust Deep Reinforcement Learning against Adversarial Behavior Manipulation](https://mlsanity.com/paper/2406.03862/) (2024)
- [On Minimizing Adversarial Counterfactual Error in Adversarial RL](https://mlsanity.com/paper/2406.04724/) (2024)
- [Robust Critics: Defending LLMs Against Multi-Turn Attacks](https://mlsanity.com/paper/2607.20472/) (2026)

## Abstract (arXiv)

Deep Reinforcement Learning (DRL) policies have been shown to be vulnerable to small adversarial noise in observations. Such adversarial noise can have disastrous consequences in safety-critical environments. For instance, a self-driving car receiving adversarially perturbed sensory observations about nearby signs (e.g., a stop sign physically altered to be perceived as a speed limit sign) or objects (e.g., cars altered to be recognized as trees) can be fatal. Existing approaches for making RL algorithms robust to an observation-perturbing adversary have focused on reactive approaches that iteratively improve against adversarial examples generated at each iteration. While such approaches have been shown to provide improvements over regular RL methods, they are reactive and can fare significantly worse if certain categories of adversarial examples are not generated during training. To that end, we pursue a more proactive approach that relies on directly optimizing a well-studied robustness measure, regret instead of expected value. We provide a principled approach that minimizes maximum regret over a "neighborhood" of observations to the received "observation". Our regret criterion can be used to modify existing value- and policy-based Deep RL methods. We demonstrate that our approaches provide a significant improvement in performance across a wide variety of benchmarks against leading approaches for robust Deep RL.