CUED Publications database

Minimum description length vs. maximum likelihood in lossy data compression

Madiman, M and Harrison, M and Kontoyiannis, I (2004) Minimum description length vs. maximum likelihood in lossy data compression. In: UNSPECIFIED 461-..

Full text not available from this repository.

Abstract

The relation between minimum description length and maximum likelihood methods are compared in lossy data compression. The starting point is the precise correspondence between compression algorithms and probability distributions Q on Ân. Lossless MLE tends to overfit the data, so as to rectify this, the penalized versions of LMLE was considered, and the Lossy Minimum Description Length Estimate (LMDLE) was defined. The results of Repeated simulations show that the LMDLE hits and stays at θ * quite fast, unlike the LMLE which bounces around forever.

Item Type: Conference or Workshop Item (UNSPECIFIED)
Subjects: UNSPECIFIED
Divisions: Div F > Signal Processing and Communications
Depositing User: Cron Job
Date Deposited: 08 Jan 2018 20:12
Last Modified: 18 Aug 2020 12:42
DOI: