Abstraction selection in model-based reinforcement learning

Keyphrases

• Approximation Error 50%
• Estimation Error 50%
• Finite-sample Guarantees 50%
• Large Amount of Data 50%
• Levels of Abstraction 50%
• Limited Quantity 50%
• Loss Bounds 50%
• Lower Approximation 50%
• Model Complexity 50%
• Model-based Reinforcement Learning 100%
• Planning Horizon 50%
• Problems in Practice 50%
• SIMPLE Algorithm 50%
• State Abstraction 50%
• Statistical Hypothesis Testing 50%

Computer Science

• Approximation (Algorithm) 50%
• Estimation Error 50%
• Model-Based Reinforcement Learning 100%
• Planning Horizon 50%
• Simple Algorithm 50%