School
// Draw agent geometry with stroke, then fill polygons
20/Aug/2008 10:45 PM
The summer is almost over. I can't believe how
quickly it went by. But I'm getting back into the
swing of things by working diligently on my long-term
research project. Read
More...
break;
15/Jul/2008 11:49 PM
Last Monday, the WSU Summer Camp started. Its been a
rather interesting experience, with a rough start, to
say the least. When you've got five pair of
adolescent eyes glazed over while you try to stress
the importance of the 16-bit video game era and the
influence therein as it pertains to the emergence of
contemporary market trends, it doesn't hurt to take a
step back and just let them play Flash games for a
while. Fuck.
Anyways, all my programming projects have been postponed, much to my own chagrin. There exists this trend, it seems, that whenever I start to gain steam, things inevitably get derailed far too premature. It would also seem that I subconsciously prefer uninventive locomotive puns to chug my points across. In any case, my BreakOUT game is on hold, and any code I planned on writing will instead need to be relevant to my research project. This isn't too bad, considering my project requires quite a lot of artificial intelligence in 3D space; that's kinda like making a video game, right?
Anyways, all my programming projects have been postponed, much to my own chagrin. There exists this trend, it seems, that whenever I start to gain steam, things inevitably get derailed far too premature. It would also seem that I subconsciously prefer uninventive locomotive puns to chug my points across. In any case, my BreakOUT game is on hold, and any code I planned on writing will instead need to be relevant to my research project. This isn't too bad, considering my project requires quite a lot of artificial intelligence in 3D space; that's kinda like making a video game, right?
void Research(const Student &kevin) {}
27/Mar/2008 03:25 PM
Research brainstorm went down today in FourO'Nine:
What I'm going to do is gather hundreds of hours of Madden'0X gameplay/footage for the purpose of teaching a computer how to beat a human opponent. I would take a human player (or players), and just observe their playcalls over and over, collecting, then logically sorting through, the information on what plays they called and under what situations, etc. I would use data mining techniques based in cultural algorithms to then, months after collecting this data, invite the player(s) back to play against a computer deciding on calls throughout the game. Decisions made by the Madden-playing framework would all come through based on, again, cultural algorithms as to which call would be best. For better results, both the human player and the framework will simply coach the game (only making calls), and the game's original AI would execute the on-field action. I feel that this research would lead to further uses of cultural algorithms to dynamically change AI agents within games to be ubiquitous to any player's style and skill level.
Madden research would be most excellent.
What I'm going to do is gather hundreds of hours of Madden'0X gameplay/footage for the purpose of teaching a computer how to beat a human opponent. I would take a human player (or players), and just observe their playcalls over and over, collecting, then logically sorting through, the information on what plays they called and under what situations, etc. I would use data mining techniques based in cultural algorithms to then, months after collecting this data, invite the player(s) back to play against a computer deciding on calls throughout the game. Decisions made by the Madden-playing framework would all come through based on, again, cultural algorithms as to which call would be best. For better results, both the human player and the framework will simply coach the game (only making calls), and the game's original AI would execute the on-field action. I feel that this research would lead to further uses of cultural algorithms to dynamically change AI agents within games to be ubiquitous to any player's style and skill level.
Madden research would be most excellent.