We gratefully acknowledge support from
the Simons Foundation and member institutions.
Full-text links:

Download:

(license)

Current browse context:

cs
< prev | next >
new | recent | 0201

References & Citations

Bookmark

(what is this?)
CiteULike logo BibSonomy logo Mendeley logo del.icio.us logo Digg logo Reddit logo

Computer Science > Machine Learning

Title: Sharpening Occam's Razor

Authors: Ming Li (Univ. Waterloo), John Tromp (CWI), Paul Vitanyi (CWI and University of Amsterdam)
(Submitted on 8 Jan 2002 (v1), last revised 10 Oct 2002 (this version, v2))
Abstract: We provide a new representation-independent formulation of Occam's razor theorem, based on Kolmogorov complexity. This new formulation allows us to:
(i) Obtain better sample complexity than both length-based and VC-based versions of Occam's razor theorem, in many applications.
(ii) Achieve a sharper reverse of Occam's razor theorem than previous work.
Specifically, we weaken the assumptions made in an earlier publication, and extend the reverse to superpolynomial running times.
Comments: LaTeX 13 pages; Proc 8th COCOON, LNCS 2387, Springer-Verlag, Berlin, 2002, 411--419
Subjects: Machine Learning (cs.LG); Disordered Systems and Neural Networks (cond-mat.dis-nn); Artificial Intelligence (cs.AI); Computational Complexity (cs.CC); Probability (math.PR); Data Analysis, Statistics and Probability (physics.data-an)
ACM classes: F.2, E.4, I.2
Report number: CWI Manuscript 1994
Cite as: arXiv:cs/0201005 [cs.LG]
  (or arXiv:cs/0201005v2 [cs.LG] for this version)

Submission history

From: Paul Vitanyi [view email]
[v1] Tue, 8 Jan 2002 16:44:10 GMT (11kb)
[v2] Thu, 10 Oct 2002 17:23:57 GMT (14kb)
Which authors of this paper are endorsers? | Disable MathJax (What is MathJax?)

Link back to: arXiv, form interface, contact.