This program shows how an Q-learning algorithm works in a maze and has a maze-editor to create own test senarios.
Developed with the help of an already existing program: http://www.cs.cmu.edu/~awm/rlsim/
The application comes with a gradle wrapper, which loads all dependencies and runs the application with the command:
sh gradlew run
Q(s,a) = Q(s,a) + α [r + γ * Q(s',a') - Q(s,a)]
Legend:
- s - state
- a - action
- s'/a' - next state/action
- r - reward/penalty
- γ - discounting rate
- α - learning rate