Corruption-Robust Offline Reinforcement Learning with General Function Approximation

Keyphrases

• Corruption 100%
• Function Approximation 100%
• General Functioning 100%
• Offline Reinforcement Learning 100%
• Markov Decision Process 42%
• Uncertainty Weighting 28%
• Tight 14%
• Regularization Parameter 14%
• Weighting Method 14%
• Robust Algorithm 14%
• Feature Map 14%
• Data Distribution 14%
• Learning Settings 14%
• Iterative Methods 14%
• Corrupt 14%
• Error Term 14%
• Optimal Policy 14%
• Adversary 14%
• Additive Factors 14%
• Confidence Set 14%
• Dependent Errors 14%
• Corruption Level 14%
• Interactive Reinforcement Learning 14%
• Corruption Robustness 14%

Mathematics

• Approximation Function 100%
• Markov Decision Process 100%
• Minimizes 33%
• Regularization 33%
• Error Term 33%
• Optimal Policy 33%
• Robust Algorithm 33%
• Data Distribution 33%
• Confidence Set 33%

Computer Science

• Reinforcement Learning 100%
• Function Approximation 100%
• Markov Decision Process 75%
• Regularization Parameter 25%
• Data Distribution 25%
• Underlying Data 25%