Reinforcement Learning

  1. Function Approximation
    1. Need for Function Approximation
      1. Large State Spaces
        1. Continuous State Spaces
          1. Curse of Dimensionality
            1. Generalization Requirements
            2. Value Function Approximation
              1. Approximate Value Functions
                1. Function Approximation Architectures
                  1. Feature Representation
                    1. Hand-Crafted Features
                      1. Basis Functions
                        1. Feature Selection
                        2. Linear Function Approximation
                          1. Linear Combinations of Features
                            1. Weight Vector Learning
                              1. Convergence Properties
                              2. Nonlinear Function Approximation
                                1. Neural Networks
                                  1. Decision Trees
                                    1. Kernel Methods
                                  2. Prediction with Function Approximation
                                    1. Gradient-Based Methods
                                      1. Stochastic Gradient Descent
                                        1. Learning Rate Schedules
                                          1. Convergence Analysis
                                          2. Semi-Gradient Methods
                                            1. Bootstrapping with Function Approximation
                                              1. Stability Issues
                                                1. Convergence Conditions
                                                2. Least-Squares Methods
                                                  1. LSTD (Least-Squares TD)
                                                    1. Computational Complexity
                                                      1. Batch Updates
                                                    2. Control with Function Approximation
                                                      1. Action-Value Function Approximation
                                                        1. Semi-Gradient SARSA
                                                          1. Semi-Gradient Q-Learning
                                                            1. Policy Gradient Methods
                                                            2. Stability and Convergence Issues
                                                              1. The Deadly Triad
                                                                1. Function Approximation
                                                                  1. Bootstrapping
                                                                    1. Off-Policy Learning
                                                                    2. Divergence Examples
                                                                      1. Stabilization Techniques
                                                                      2. Feature Construction
                                                                        1. Tile Coding
                                                                          1. Radial Basis Functions
                                                                            1. Fourier Basis
                                                                              1. Polynomial Features

                                                                            Previous

                                                                            6. Temporal-Difference Learning

                                                                            Go to top

                                                                            Next

                                                                            8. Deep Reinforcement Learning