RORL: Robust Offline Reinforcement Learning via Conservative Smoothing
Explore this paper's citation graph
Summary
This work proposes Robust Offline Reinforcement Learning (RORL), a novel conservative smoothing technique that can achieve state-of-the-art performance on the general offline RL benchmark and is considerably robust to adversarial observation perturbations.
- Type
- preprint
- Published
- 2022-06-06
- Cited by
- 119
- References
- 82
- Access
- Open access
- OpenAlex
- https://openalex.org/W4281794919
- Semantic Scholar
- https://api.semanticscholar.org/CorpusID:249431425
Keywords
Reinforcement learning, Robustness (evolution), Computer science, Smoothing, Conservatism
References
- Improved Algorithms for Linear Stochastic Bandits
- Probabilistic Principal Component Analysis
- Human-level control through deep reinforcement learning
- Robust Dynamic Programming
- Robustness in Markov Decision Problems with Uncertain Transition Matrices
- Scaling Up Robust MDPs by Reinforcement Learning
- Mastering the game of Go with deep neural networks and tree search
- Practical Black-Box Attacks against Deep Learning Systems using Adversarial Examples
- Robust H/sub infinity / control for linear systems with norm-bounded time-varying uncertainty
- Vulnerability of Deep Reinforcement Learning to Policy Induction Attacks
- Adversarial Attacks on Neural Network Policies
- Minimax Regret Bounds for Reinforcement Learning
- Domain randomization for transferring deep neural networks from simulation to the real world
- Fast Bellman Updates for Robust MDPs
- Exponentially Weighted Imitation Learning for Batched Historical Data
- Active Domain Randomization
- Stabilizing Off-Policy Q-Learning via Bootstrapping Error Reduction
- Provably Efficient Reinforcement Learning with Linear Function Approximation
- Robust Deep Reinforcement Learning with Adversarial Attacks
- Mastering Atari, Go, chess and shogi by planning with a learned model
Cited by
- What is Flagged in Uncertainty Quantification? Latent Density Models for Uncertainty Categorization
- First-order Policy Optimization for Robust Markov Decision Process
- Robust Offline Reinforcement Learning with Gradient Penalty and Constraint Relaxation
- Monotonic Quantile Network for Worst-Case Offline Reinforcement Learning
- Offline Reinforcement Learning with Closed-Form Policy Improvement Operators
- One Risk to Rule Them All: A Risk-Sensitive Perspective on Model-Based Offline Reinforcement Learning
- Anti-Exploration by Random Network Distillation
- Behavior Proximal Policy Optimization
- Recover Triggered States: Protect Model Against Backdoor Attack in Reinforcement Learning
- Revisiting the Minimalist Approach to Offline Reinforcement Learning
- What is Essential for Unseen Goal Generalization of Offline Goal-conditioned RL?
- Survival Instinct in Offline Reinforcement Learning
- Policy Regularization with Dataset Constraint for Offline Reinforcement Learning
- ENOTO: Improving Offline-to-Online Reinforcement Learning with Q-Ensembles
- Robust Offline Reinforcement Learning - Certify the Confidence Interval
- Accountability in Offline Reinforcement Learning: Explaining Decisions with a Corpus of Examples
- Towards Robust Offline Reinforcement Learning under Diverse Data Corruption
- Corruption-Robust Offline Reinforcement Learning with General Function Approximation
- GOPlan: Goal-conditioned Offline Reinforcement Learning by Planning with Learned Models
- Balancing policy constraint and ensemble size in uncertainty-based offline reinforcement learning
Related papers
- Analysis of American Conservatism in 1960s
- The Conservative Tradition in America
- American History Analyzed by Conservatism——A Macroscopic Investigation on American History
- The Promise and Danger of National Conservatism
- Liberal or conservative? Gender, identity, and perception of historical religious positions.
- On the Neo-Conservatism of U.S.A
- Modern Conservatism: The Problem of Definition