Jürgen Schmidhuber

Technische Universität München (TUM), Germany

Istituto Dalle Molle di Studi sull'Intelligenza Artificiale (IDSIA), Switzerland

I argue that visual and other data becomes temporarily interesting by itself to a self-improving, but
computationally limited, subjective observer once he learns to predict or compress the data in a better
way, thus making it subjectively simpler and more "beautiful." Curiosity is the desire to create or
discover more non-random, non-arbitrary, regular data that is novel and surprising not in the traditional
sense of Boltzmann and Shannon but in the sense that it allows for compression progress because its
regularity was not yet known. This drive maximizes interestingness, the first derivative of subjective
beauty or compressibility, that is, the steepness of the learning curve. It motivates visual and other types
of attention in exploring infants, pure mathematicians, composers, artists, dancers, comedians, yourself,
and recent artificial systems. 

Based on keynote for KES 2008 and joint invited lecture for ALT 2007 / DS 2007; variants to appear in SICE Journal & Proc. ABIALS.

arXiv preprint: http://arXiv.org/abs/0812.4360

WWW sites on this topic: http://www.idsia.ch/~juergen/interest.html
and also http://www.idsia.ch/~juergen/beauty.html