Fordham RL Tutorial 2019
later has come.
Reinforcement learning is a powerful paradigm by which artificially intelligent agents can learn to make good decisions. RL is relevant to an enormous range of tasks, including robotics, game playing, consumer modeling and healthcare. This class will provide a solid introduction to the field of reinforcement learning and students will learn about the core challenges and approaches, including generalization and exploration. Students will become well versed in key ideas and techniques for RL. The class will be project oriented. Assignments will include the basics of reinforcement learning as well as imitation learning and deep reinforcement learning. In addition, students will advance their understanding and the field of RL through an open-ended class project.
0. Tic-Tac-Toe 1. Multi-armed Bandits 2. Exploration/Exploitation 3. Markov Processes - Sutton - Russel & Norvig 4. Prediction - Estimating Value Functions 5. Control - Monte Carlo - Temporal Difference - SARSA - Q Learning 6. Planning - n-Step SARSA - DYNA-Q - Monte Carlo Tree Search 7. Deep Q Learning - Achieving Human level control in Atari Games -- Mnih et al 8. Policy Gradient Methods - Actor Critic 9. Imitation Learning - Behaviour Cloning - Inverse Reinforcement Learning - Latent Variable Models - Reframing Control as Probablistic Inference - IRL - Margin Based <-- introduced - Entropy Based <-- focused on - GANs, IRL, and Energy Based Models -- Finn et al - Imitation Learning with Concurrent action in 3D Games -- Harmer et al
Sadly, I cannot recall where I found the picture of Baymax from. Credit is not mine on that.