My implementation of machine learning for minesweeper solver
Modified Q-Understanding implementation from this write-up (http://cs229.stanford.edu/proj2015/372_report.pdf)
The long run reward part of the equation is dismissed as I am not interested in conclusion game, but relatively in rapid reward for current state.
Built in Unity 2018.2
I save states in scriptable item, any time I tried to help save far more than 3000000 unity would get up to 7 GB of ram usage and crash, so I desired to divided into quite a few scriptables with fifty percent million documents every single.The game titles in which a new condition was found out ended up disregard in the win charge calculation.
the info occupies more than 1.1GB on the disc and it usually takes a minute or so to start the venture :D.
The solver doesnt established the flags, the activity is deemed a gain when all blocks apart from the bombs are uncovered.
Also this is initial time I ever recorded voice so remember to be being familiar with.
