Metrics for symbol clustering from a pseudoergodic information source
Ángel Kuri-Morales, Oscar Herrera-Alcántara · 2004
We discuss a set of metrics, which aim to facilitate the formation of symbol groups from a pseudoergodic information source. An optimal codification can then be applied on the symbols (such as Huffman Codes by S. JR Pierce (1980)) for aero memory sources where it tends to the theoretical limit of compression limited by the entropy. These metrics can be used as a fitness measure of the individuals in the Vasconcelos genetic algorithm as an alternative to exhaustive search.