At 11/6/11 07:13 PM, MSGhero wrote:
Any of you guys have experience with Q learning or SARSA or some other reinforcement learning algorithm? I'm looking into it for a new project, but I don't quite understand "states." Like how general does the state have to be or what should I include in it. It's gonna be a fighting game, maybe with combos and stuff.
Q learning is somewhere in this article...
What a coincidence, the AI Class's unit this week just so happens to be "Reinforcement learning" which includes Q learning. I'm not sure if the videos are available on this to those who aren't enrolled, but you can watch them on youtube anyway.
Speaking of that AI class, who else here is following? I remember Archon saying he'd signed up. How you following so far? I think I'm following well, and everything's good in theory, but what I'm not really getting the hang of is how I can use these concepts to solve practical problems or just use them in game AI or anything so far.