Guarantees for Epsilon-Greedy Reinforcement Learning with Function Approximation

Keyphrases

• Value Function 100%
• Sample Complexity 100%
• Reinforcement Learning 100%
• Function Approximation 100%
• Exploration Policy 100%
• Epsilon-greedy 100%
• Regret 50%
• Class Function 50%
• Dynamic Structure 50%
• Structural Properties 50%
• Complexity Bounds 50%
• Function-based 50%
• Complexity Measures 50%
• Gaussian Noise 50%
• Reinforcement Learning Task 50%
• Reward Structure 50%
• Softmax 50%
• Bellman 50%
• Episodic MDP 50%

Computer Science

• Reinforcement Learning 100%
• Function Approximation 100%
• Function Value 66%
• Structural Property 33%
• Dynamic Structure 33%
• Complexity Measure 33%
• Gaussian White Noise 33%
• Concrete Example 33%
• Reward Structure 33%

Mathematics

• Function Value 100%
• Approximation Function 100%
• Gaussian Distribution 50%
• Structural Property 50%
• Learning Task 50%