I search for solutions in this order: Past Code, Unreal Source, Wiki, BUF, groups.yahoo, google, screaming at monitor. – RegularX
Legacy:Juxtapose
Juxtapose[edit]
I'm trying to do a project with AI in Unreal for my Machine Learning Class. This is my first attempt at doing anything with Unreal Code after playing UT since it came out. I've posted to the wiki a few times anonymously and I can't find them. If anyone knows what happened to my earlier edits I'd be much obliged to hear about it.
Learning Onslaught[edit]
Not much to say here. It's not working and internally still very screwed up.
The goal is to create an AI that learns from its actions based upon what happens shortly after the actions are taken. The game restricts the teams to five equally skilled bots. The Learning AI coordinates bot actions by prioritizing targets for squads. When the state of the game changes (a node gets build, one team loses e.g.) the game rewards or punishes the AI accordingly (before you ask, its with arbitrary values).
If anyone is further interested and acquainted with ML, I'm using Q-Learning so I don't need a model and calculating the state space across the status of individual nodes.
