Preprint, ResearchSquare, posted September 12, 2022. Last accessed September 20, 2022.
ISSN/ISBN: Not available at this time. DOI: 10.21203/rs.3.rs-2042757/v1
Abstract: The question of efficient coding of information for engineered and natural systems is connected both with the use of applicable probability distribution and assignment of appropriate dimensionality with the data. When the cost of coding of data or that of classes increases linearly with the number of bases, ternary coding is superior to binary, and coding in e is optimal. This paper investigates the relative efficiency of bases for the cases when the cost complexity is affine (slope-intercept linear) or exponential. The affine cases for which binary and ternary bases are optimal are presented. Noninteger number of bases imply self-similarity and, therefore, this study forms a bridge to the use of the Newcomb-Benford distribution in a variety of applications.
Bibtex:
@misc{,
author = {Subhash, Kak},
year = {2022},
title = {Coding Complexity Cost and the Optimal Number of Bases},
doi = {10.21203/rs.3.rs-2042757/v1},
url = {https://www.researchsquare.com/article/rs-2042757/v1},
}
Reference Type: Preprint
Subject Area(s): Computer Science