I would like to use neuronal networks and related concepts with Julia. The idea is to start with Tic-Tac-Toe, then look at more general m,n,k-games. I would like to be flexible for experiments with different networks, size and number of hidden layers etc.
A simple way to represent the board is using an m * n matrix with +1, -1, 0 for “X”, “O”, “empty cell”. Therefore the network should have the m * n input layer, at least one hidden layer and the m * n output layer indicating the probability of the next move per cell.
I had a look at the available resources in https://juliapackages.com/c/ai
Which package do you propose to start with? What about ease of use and performance? (I don’t need a GPU backend)