An improved sequential search multistage vector quantizer
- 31 December 2002
- conference paper
- Published by Institute of Electrical and Electronics Engineers (IEEE)
Abstract
A new structure permits improved solutions which approximate the exhaustive-search multistage solution. A deterministic annealing design method capitalizing on this structure is formulated within the framework of information theory. The sequential search constraint is included as a prior, and the principal of minimum cross entropy is invoked. The method obtains improvement over both the standard sequential design and joint optimization approaches.<>Keywords
This publication has 17 references indexed in Scilit:
- Multiple stage vector quantization for speech codingPublished by Institute of Electrical and Electronics Engineers (IEEE) ,2005
- Vector quantization by deterministic annealingIEEE Transactions on Information Theory, 1992
- Vector Quantization and Signal CompressionPublished by Springer Nature ,1992
- Parallel and deterministic algorithms from MRFs: surface reconstructionPublished by Institute of Electrical and Electronics Engineers (IEEE) ,1991
- A deterministic annealing approach to clusteringPattern Recognition Letters, 1990
- Statistical mechanics and phase transitions in clusteringPhysical Review Letters, 1990
- Generalized Deformable Models, Statistical Physics, and Matching ProblemsNeural Computation, 1990
- Stochastic Relaxation, Gibbs Distributions, and the Bayesian Restoration of ImagesPublished by Institute of Electrical and Electronics Engineers (IEEE) ,1984
- Optimization by Simulated AnnealingScience, 1983
- Axiomatic derivation of the principle of maximum entropy and the principle of minimum cross-entropyIEEE Transactions on Information Theory, 1980