How many ways can AI play tic tac toe?
I am doing this to learn, tic tac toe is one of the simplest games so my goal is to learn many algorithms by implementing them with tic tac toe, connect four is one of my favourite games so that's probably next too (that's why there's a rogue minimax +AB-pruning connect four dir)
Here's the roadmap:
minimax and alpha-beta pruning Montecarlo tree search RL based neural network use object detection so you play on paper Genetic Algorithms - Represent Tic Tac Toe strategies as sets of rules or heuristics.
Also Muzero would be cool
OTHER NOTES:
After I do this stuff I want to do MNIST number detection from scratch, want to start with pytorch, then maybe try tensorflow to learn about that, but also from only numpy would be cool like geohot, and can try many other object detection algos, CNN algos.