On the Value of Interaction and Function Approximation in Imitation Learning

Keyphrases

• Function Approximation 100%
• Imitation Learning 100%
• Recoverability 40%
• Learning Problems 20%
• Optimal Algorithm 20%
• High Probability 20%
• Learning Algorithm 20%
• Sampling Efficiency 20%
• Statistical Guarantees 20%
• Transition Structure 20%
• Optimal Policy 20%
• Linear Classifier 20%
• Multi-class Classification 20%
• Information-theoretic Bounds 20%
• State Action 20%
• Linear Function Approximation 20%
• Confidence Set 20%
• Minimax Rate 20%
• Action Features 20%
• Linear Classification 20%
• Parameter Sharing 20%
• Behavioral Cloning 20%
• Episodic MDP 20%
• Play Actions 20%
• Expert Trajectories 20%

Mathematics

• Approximation Function 100%
• Probability Theory 50%
• Time Step 50%
• Minimax 50%
• Worst Case 50%
• Linear Function 50%
• Optimal Policy 50%
• Confidence Set 50%

Computer Science

• Function Approximation 100%
• Optimal Algorithm 33%
• Learning Algorithm 33%
• Learning Problem 33%
• Linear Classifier 33%
• Multiclass Classification 33%
• Multi Class Classification 33%
• Linear Function 33%