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

Download:

Current browse context:

cs.LG
< prev | next >
new | recent | 1005

Change to browse by:

cs
cs.CC

References & Citations

DBLP - CS Bibliography

listing | bibtex

Bookmark

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

Computer Science > Machine Learning

Title: A Short Introduction to Model Selection, Kolmogorov Complexity and Minimum Description Length (MDL)

Authors: Volker Nannen
(Submitted on 13 May 2010 (v1), last revised 14 May 2010 (this version, v2))
Abstract: The concept of overfitting in model selection is explained and demonstrated with an example. After providing some background information on information theory and Kolmogorov complexity, we provide a short explanation of Minimum Description Length and error minimization. We conclude with a discussion of the typical features of overfitting in model selection.
Comments: 20 pages, Chapter 1 of The Paradox of Overfitting, Master's thesis, Rijksuniversiteit Groningen, 2003
Subjects: Machine Learning (cs.LG); Computational Complexity (cs.CC)
MSC classes: F.2.3
Cite as: arXiv:1005.2364 [cs.LG]
  (or arXiv:1005.2364v2 [cs.LG] for this version)

Submission history

From: Volker Nannen [view email]
[v1] Thu, 13 May 2010 15:59:01 GMT (78kb,D)
[v2] Fri, 14 May 2010 11:28:03 GMT (78kb,D)
Which authors of this paper are endorsers? | Disable MathJax (What is MathJax?)

Link back to: arXiv, form interface, contact.