Pitfall-2.5x.mov

Andre Cohen, Carlos Diuk, and Michael Littman (Rutgers University) designed a learning algorithm that discovers how to get off the first screen of the classic Atari game Pitfall. During the first run, it is experimenting with different actions and in the second run it uses what it learned to exit the screen. Cue happy dance.

Category:

Education

Tags:

• reinforcement learning
• atari
• pitfall
• artificial intelligence

[ThePrussianGangsta]

at the very end he demonstrates he can celebrate!

i cant help but imagine the terminator doing that dance after he kills john Connor.

[Demiath]

Oh great, we're all gonna die now.

[alextsiri]

@xeriqx lol

[wberdan]

this is some horrible playing.

[sunmanpatoo]

u should program him to collaborate with neyo in the matrix, then they both can save zion!

[EnderDio]

Amazing! Video Game AI! What a new and wonderful idea!

this comment is a joke.

[xeriqx]

Skynet started like this.... like a Chess PC player! ( i think so, i watched it on Sarah Connor Chronicles

[asphyxiateDrake00]

We need to not make learning algorithms like this or I fear that a robot apocalypse will most definitely ensue.

[gmourao]

omg...MARKOV SUCKS!