Entropy, Derivation Operators and Huffman Trees

Simon Burton
DOI: https://doi.org/10.48550/arXiv.2107.11208
2021-07-23
Abstract:We build a theory of binary trees on finite multisets that categorifies, or operationalizes, the entropy of a finite probability distribution. Multisets operationalize probabilities as the event outcomes of an experiment. Huffman trees operationalize the entropy of the distribution of these events. We show how the derivation property of the entropy of a joint distribution lifts to Huffman trees.
Information Theory
What problem does this paper attempt to address?