Context Weighting for General Finite-Context Sources (1996) [10 citations — 0 self]
Popular Tags
No tags have been applied to this document.
BibTeX | Add To MetaCart
author = {Frans M. J. Willems and Yuri M. Shtarkov and Tjalling J. Tjalkens},
title = {Context Weighting for General Finite-Context Sources},
journal = {IEEE Trans. Inform. Theory},
year = {1996},
volume = {42},
pages = {42--1514}
}
Bookmarks
OpenURL
Abstract:
Context weighting procedures are presented for sources with models (structures) in four different classes. Although the procedures are designed for universal data compression purposes, their generality allows application in the area of classification. 1 Introduction Recently in [14],[15] the authors introduced context-tree weighting as a sequential universal source coding method for the class of binary (bounded memory) tree sources. Tree sources were defined around the same time by Weinberger et al. [13]. The idea behind context weighting procedures can be summarized as follows 1 : The well known Elias algorithm (described in e.g. Jelinek [1]) produces for any coding distribution P c (x T 1 ) over all binary sequences of length T , a binary prefix code with codeword lengths L(x T 1 ) that satisfy L(x T 1 ) ! log 1 P c (x T 1 ) + 2 for all x T 1 : (1) (We assume that the base of the log(\Delta) is 2. Codeword lengths and information quantities are expressed in bits.) If th...
