Closed hapu closed 5 years ago
we should look for more reinforcement learning approaches
A popular one is NEAT (Neuroevolution of Augmenting topologies) https://www.youtube.com/watch?v=qv6UVOQ0F44
The original white paper is: http://nn.cs.utexas.edu/downloads/papers/stanley.ec02.pdf
what about a mini game of tictactoe where every time you beat the machine the machine learns a new way to win