Title :
Non-uniform PPM and context tree models
Author :
Åberg, J. ; Shtarkov, Yu.M. ; Smeets, B.J.M.
Author_Institution :
Dept. of Inf. Technol., Lund Univ., Sweden
fDate :
30 Mar-1 Apr 1998
Abstract :
The problem of optimizing PPM with the help of different choices of estimators and their parameters for different subsets of nodes in the context tree is considered. Methods of such optimization for Markov chain and context tree models for individual files and over given sets of files are presented, and it is demonstrated that the extension from Markov chain models to context tree models is necessary to receive significant improvements of the compression ratio
Keywords :
Markov processes; data compression; optimisation; parameter estimation; trees (mathematics); Markov chain models; compression ratio; context tree models; estimators; files; nodes; nonuniform PPM; optimization; Algorithm design and analysis; Context modeling; Information technology; Information theory; Optimization methods; Parameter estimation; Probability;
Conference_Titel :
Data Compression Conference, 1998. DCC '98. Proceedings
Conference_Location :
Snowbird, UT
Print_ISBN :
0-8186-8406-2
DOI :
10.1109/DCC.1998.672156
