I'm working in game dev, and on side project I'm currently building a little (newbie) ml framework in c# to discover FFNN, CNN and probly RNN. I'm currently struggling on the backpropagation in convolutionnal layer but that's a matter of time before it works (I hope) 😂

I'm very curious to see the possible applications in game AI. I already did some testing projects before, simple ML agents with small fully connected networks.. But I want to go further and probly try to mix the utility based ai pattern with reinforcement learning methods or genetic algorithm.

I also think Convolutionnal network could maybe be used to input some 'spatialized' data to an ai agent and allow him to take decision about movement or so..