On the Representative Power of Commented Markov Models

TitleOn the Representative Power of Commented Markov Models
Publication TypeTechnical Report
Year of Publication1996
AuthorsBlasig, R., & Fahner G.
Other Numbers1044
Abstract

A CMM (Commented Markov Model) is a learning algorithm to model and extrapolate discrete sequences. The learning involves the inferences of objects, variables and classes, describing the sequences. In this paper, all sequences considered will be character sequences. As pointed out in an earlier paper [2], the structures utilized by CMM are powerful enough to represent and evaluate any primitive recursive function. This paper will provide a formal proof of this claim. We will therefore concentrate on the issues of representation and leave the issues of CMM induction aside.

URLhttp://www.icsi.berkeley.edu/ftp/global/pub/techreports/1996/tr-96-034.pdf
Bibliographic Notes

ICSI Technical Report TR-96-034

Abbreviated Authors

R. Blasig and G. Fahner

ICSI Publication Type

Technical Report