Completed
Minecraft World Experiments
Class Central Classrooms beta
YouTube videos curated by Class Central.
Classroom Contents
Formal Languages and Automata for Reward Function Specification and Efficient Reinforcement Learning
Automatically move to the next video in the Classroom when playback concludes
- 1 Intro
- 2 Acknowledgements
- 3 Reinforcement Learning (RL)
- 4 Challenges of Real-World RL
- 5 Goals and Preferences
- 6 Linear Temporal Logic (LTL) A compelling logic to express temporal properties of traces.
- 7 Challenges to RL
- 8 Toy Problem Disclaimer
- 9 Running Example
- 10 Decoupling Transition and Reward Functions
- 11 The Rest of the Talk
- 12 Define a Reward Function using a Reward Machine
- 13 Reward Function Vocabulary
- 14 Simple Reward Machine
- 15 Reward Machines in Action
- 16 Other Reward Machines
- 17 Q-Learning Baseline
- 18 Option-Based Hierarchical RL (HRL)
- 19 HRL with RM-Based Pruning (HRL-RM)
- 20 HRL Methods Can Find Suboptimal Policies
- 21 Q-Learning for Reward Machines (QRM)
- 22 QRM In Action
- 23 Recall: Methods for Exploiting RM Structure
- 24 5. QRM + Reward Shaping (QRM + RS)
- 25 Test Domains
- 26 Test in Discrete Domains
- 27 Office World Experiments
- 28 Minecraft World Experiments
- 29 Function Approximation with QRM
- 30 Water World Experiments
- 31 Creating Reward Machines
- 32 Reward Specification: one size does not fit all
- 33 1. Construct Reward Machine from Formal Languages
- 34 Generate RM using a Symbolic Planner
- 35 Learn RMs for Partially-Observable RL
