Q* Approximation Schemes for Batch Reinforcement Learning: A Theoretical Comparison

Keyphrases

• Approximation Scheme 100%
• Theoretical Comparison 100%
• Batch Reinforcement Learning 100%
• Bellman Error 100%
• Error Estimates 50%
• Performance Loss 50%
• Weight Correction 50%
• Double Sampling 50%
• Error Propagation 50%
• Stationary Policy 50%
• Classical Algorithm 50%
• Squared Loss 50%
• Batch Data 50%
• Fitted Q-iteration 50%
• Importance Weighting 50%
• Output Stationary 50%
• Performance Guarantee 50%
• Iterative Method 50%
• Algorithm Use 50%

Computer Science

• approximation scheme 100%
• Reinforcement Learning 100%
• Q Approximation 100%
• Performance Guarantee 50%
• Error Estimation 50%
• Performance Loss 50%

Earth and Planetary Sciences

• Error Propagation 100%