Miscellaneous public code.
Simple discrete, deterministic Q learning.
Can be run in command line, or in notebook (e.g. colab).
The implemented environment example is from Tom Mitchell's machine learning textbook (1997). See http://www.cs.cmu.edu/~tom/mlbook.html .
- Environment: All actions in the goal state G loop back, with reward 0.
- Optimal V(s):
- Optimal Q(s,a):
Note: V(s) and Q(s,a) assume a discount rate of 0.9.