Prague Stringology Conference 2023

Yoav Gross, Shmuel Tomi Klein, Elina Opalinsky and Dana Shapira

Selective Weighted Adaptive Coding

Motivated by improving the processing time of the weighted compressor introduced recently, while only marginally affecting its compression performance, we propose a selective-weighted method that restricts the model based positions and/or the times the model gets updated. That is, the adaptive statistical weight oriented method combines the ability to assign higher priority to symbols that are closer to those being encoded, while extending the model to the probability distribution of symbols that are not necessarily located in consecutive positions. Several variants for selecting the representative positions for determining the model are suggested, and all are empirically shown to fulfill both objectives for small skips.

Download paper: Article in PostScript Article in PDF BibTeX Reference
 PostScript   PDF   BibTeX reference