DMala |
Sleep is overrated
|
|
|
Reged: 05/09/05
|
Posts: 3989
|
Loc: Waltham, MA
|
|
Send PM
|
|
Re: Okay Debbi Downer...
04/23/13 07:27 AM
|
|
|
> > I agree he's a douche...
On a scale from Mr. Rogers to Tucker Max he rates a Charlie Sheen. ![](//www.mameworld.info/ubbthreads/images/graemlins/awe.gif) > I also agree that it is entertaining and a novel approach to writing a game playing > "A.I.".
I think this misses the point a little. It's not about writing an AI to play a game, clearly there are much more effective ways to do that. This paper is about writing a program that can *learn* to play a game, which is a rather different and much more difficult task. You give it some sample data, just to teach it which bits are the ones that it wants to increase, and then it tries to figure out the best combination of inputs to achieve that goal. That's part of the reason why the criteria for "better" are so simplified. It's relatively easy for a program to compare memory states and determine whether certain bits went up or not. Doing something like trying to parse the video output the way a human would, would be insanely complex to the point of being impossible. It turns out humans can do an amazing amount of processing, particularly in the visual realm, with basically no conscious thought at all. What's interesting about this paper to me is that it has implications beyond just playing video games. Machine learning is kind of a hot topic right now, as it has applications for things like making suggestions based on past behavior, a la Netflix or Amazon.
|
|