A machine learning implementation of 9 men's morris

This is done as a learning exercise for machine learning (I'm not a student tho) and to have an opponent to play with :)

Since this is a fairly simple game (but more complex than tic-tac-toe), I thought it would be a good candidate to try, for the first time, to implement a machine learning model.

For each state (of the board/game), it picks randomly (weighted) from all the possible moves and keeps a list of all the moves done.

If the model wins, it will increment the weight of the moves it picked. (I guess that's the "reinforced" part of the learning)

For storage, it uses unqlite, with the state of the board (piece placement and who-to-move) as key, with the value being the (variadic) list of weighted possible moves.

To run the learning, simply pit two models against each other:

$ ./learn models/a.db models/b.db

I usually keep 4 models and pit every one at the other in a round robin fashion.

TODO: implement a way to play against the model