An Open Problem Regarding the Convergence of Universal A Priori Probability

Author:	Marcus Hutter (2002-2003)
Comments:	2 pages
Subj-class:	Probability Theory; Complexity; Learning
Reference:	Proceedings of the 16th Annual Conference on Learning Theory (COLT 2003) pages 738-740
Paper:	LaTeX  -  PostScript  -  PDF  -  Html/Gif
Slides:	PostScript - PDF

Keywords: Sequence prediction; Algorithmic Information Theory; Solomonoff's prior; universal probability; posterior convergence; Martin-Löf randomness.

Abstract: Is the textbook result that Solomonoff's universal posterior converges to the true posterior for all Martin-Löf random sequences true?

BibTeX Entry

@InProceedings{Hutter:03mlconv,
  author =       "M. Hutter",
  title =        "An Open Problem Regarding the Convergence of Universal A Priori Probability",
  booktitle =    "Proceedings of the 16th Annual Conference on Learning Theory ({COLT-2003})",
  series =       "Lecture Notes in Artificial Intelligence",
  volume =       "2777",
  editor =       "B. Sch{\"o}lkopf and M. K. Warmuth",
  publisher =    "Springer",
  address =      "Berlin",
  pages =        "738--740",
  year =         "2003",
  url =          "http://www.hutter1.net/ai/mlconv.htm",
  keywords =     "Sequence prediction; Algorithmic Information Theory;
                  Solomonoff's prior; universal probability;
                  posterior convergence; Martin-L{\"o}f randomness.",
  abstract =     "Is the textbook result that Solomonoff's universal
                  posterior converges to the true posterior for all Martin-L{\"o}f
                  random sequences true?",
}