Formal Theory of Fun & Creativity
published: Dec. 13, 2010, recorded: September 2010, views: 17537
Report a problem or upload filesIf you have found a problem with this lecture or would like to send us extra material, articles, exercises, etc., please use our ticket system to describe your request and upload the data.
Enter your e-mail into the 'Cc' field, and we will keep you updated with your request's status.
To build a creative agent that never stops generating non-trivial & novel & surprising data, we need two learning modules: (1) an adaptive predictor or compressor or model of the growing data history as the agent is interacting with its environment, and (2) a general reinforcement learner. The LEARNING PROGRESS of (1) is the FUN or intrinsic reward of (2). That is, (2) is motivated to invent interesting things that (1) does not yet know but can easily learn. To maximize expected reward, in the absence of external reward (2) will create more and more complex behaviors that yield temporarily surprising (but eventually boring) patterns that make (1) quickly improve. We discuss how this principle explains science & art & music & humor, and how to scale up previous toy implementations of the theory since 1991, using recent powerful methods for (1) prediction and (2) reinforcement learning.
Download slides: ecmlpkdd2010_schmidhuber_ftf_01.pdf (13.9 MB)
Link this pageWould you like to put a link to this lecture on your homepage?
Go ahead! Copy the HTML snippet !
Reviews and comments:
What's with the hat? All it does it focus people's attention on the fact that you must be bald...
This is fantastic stuff. The first general algorithmic theory of scientific and artistic creativity, based on maximizing a reward function that measures "fun" obtained by making new non-trivial patterns with previously unknown regularities. Almost frightening to imagine what might happen if this stuff gets implemented on humanoid robots. They will want to have fun, too, and pursue their own "creative" goals all the time, not necessarily the goals of their teachers ...
Dr. Schmidhuber's theory of humor views jokes as subjectively unexpected patterns in the sense of algorithmic information theory. After learning or understanding a new joke one needs fewer bits to encode it than before. To evaluate the compression progress or fun, just count the saved bits. I feel this is a simple yet very deep insight. And he's applying the theory of humor well throughout the talk :-)
too primitive, disputable and useless theory. author has to study more abstract mathematics, he seems to be bad in it :). cap is funny, it seems to be a part of image or as it was told substitution of hair :-D
Erick S.K. writing letters again ... if you found a flaw in the math, which I greatly doubt, why don't you publish your findings in a peer-reviewed journal? You could submit your analysis to the IEEE Transactions, which published a survey of the theory.
Write your own review or comment: