B O S T O N   U N I V E R S I T Y
                                       
                      Computer Science Department
                                       
                         C O L L O Q U I U M

      Towards a Universal Theory of Artificial Intelligence based on
           Algorithmic Probability and Sequential Decision Theory

                            Marcus Hutter

                          IDSIA Switzerland

                            Monday, June 25
			        4:00 pm
                       (Coffee served at 3:45 pm)
                         Seminar Room / MCS 135

Decision theory formally solves the problem of rational agents in
uncertain worlds if the true environmental prior probability
distribution is known. Solomonoff's theory of universal induction
formally solves the problem of sequence prediction for unknown
prior distribution. We combine both ideas and get a parameterless
theory of universal Artificial Intelligence. We give strong
arguments that the resulting AI$\xi$ model is the most intelligent
unbiased agent possible. We outline for a number of problem
classes, including sequence prediction, strategic games, function
minimization, reinforcement and supervised learning, how the
AI$\xi$ model can formally solve them. The major drawback of the
AI$\xi$ model is that it is uncomputable. To overcome this
problem, we construct a modified algorithm AI$\xi^{tl}$, which is
still effectively more intelligent than any other time $t$ and
space $l$ bounded agent. The computation time of AI$\xi^{tl}$
is of the order $t\!\cdot\!2^l$. Other discussed topics are formal
definitions of intelligence order relations, the horizon problem
and relations of the AI$\xi$ theory to other AI approaches.

ftp://ftp.idsia.ch/pub/techrep/IDSIA-14-00.ps.gz

Host: Peter Gacs
-------------------------------------------------------------------------------
For colloquium info, including directions, see http://cs-www.bu.edu/colloquium
-------------------------------------------------------------------------------