Slides in PDF - of Marcus Hutter
Slides in PDF - of Marcus Hutter
Slides in PDF - of Marcus Hutter
Create successful ePaper yourself
Turn your PDF publications into a flip-book with our unique Google optimized e-Paper software.
<strong>Marcus</strong> <strong>Hutter</strong> - 76 - Universal Induction & Intelligence<br />
A universal choice <strong>of</strong> ξ and M<br />
• We have to assume the existence <strong>of</strong> some structure on the<br />
environment to avoid the No-Free-Lunch Theorems [Wolpert 96].<br />
• We can only unravel effective structures which are describable by<br />
(semi)computable probability distributions.<br />
• So we may <strong>in</strong>clude all (semi)computable (semi)distributions <strong>in</strong> M.<br />
• Occam’s razor and Epicurus’ pr<strong>in</strong>ciple <strong>of</strong> multiple explanations tell<br />
us to assign high prior belief to simple environments.<br />
• Us<strong>in</strong>g Kolmogorov’s universal complexity measure K(ν) for<br />
environments ν one should set w ν ∼ 2 −K(ν) , where K(ν) is the<br />
length <strong>of</strong> the shortest program on a universal TM comput<strong>in</strong>g ν.<br />
• The result<strong>in</strong>g AIXI model [<strong>Hutter</strong>:00] is a unification <strong>of</strong> (Bellman’s)<br />
sequential decision and Solomon<strong>of</strong>f’s universal <strong>in</strong>duction theory.