### Abstract

In this paper, we develop a deterministic annealing approach for the design of entropy-constrained vector quantizers. Here, the quantization problem is reformulated within a probabilistic setting and the principle of maximum entropy is invoked. The resulting method is connected to statistical physics in two important respects. First, there is annealing in the process controlled by a `temperature' parameter, useful for avoiding local optima of the cost. Second, the number of distinct codevectors grows by a sequence of phase transitions as the temperature is varied. When the temperature is lowered to zero, the method searches for the optimal entropy-constrained vector quantizer. Conditions for bifurcation in the process provide insights into the `right' codebook size for a given problem.

Original language | English (US) |
---|---|

Title of host publication | Conference Record of the Asilomar Conference of Signals, Systems & Computers |

Publisher | Publ by IEEE |

Pages | 1651-1655 |

Number of pages | 5 |

ISBN (Print) | 0818641207 |

State | Published - Dec 1 1993 |

Event | Proceedings of the 27th Asilomar Conference on Signals, Systems & Computers - Pacific Grove, CA, USA Duration: Nov 1 1993 → Nov 3 1993 |

### Publication series

Name | Conference Record of the Asilomar Conference of Signals, Systems & Computers |
---|---|

Volume | 2 |

ISSN (Print) | 1058-6393 |

### Other

Other | Proceedings of the 27th Asilomar Conference on Signals, Systems & Computers |
---|---|

City | Pacific Grove, CA, USA |

Period | 11/1/93 → 11/3/93 |

### Fingerprint

### All Science Journal Classification (ASJC) codes

- Signal Processing
- Computer Networks and Communications

### Cite this

*Conference Record of the Asilomar Conference of Signals, Systems & Computers*(pp. 1651-1655). (Conference Record of the Asilomar Conference of Signals, Systems & Computers; Vol. 2). Publ by IEEE.

}

*Conference Record of the Asilomar Conference of Signals, Systems & Computers.*Conference Record of the Asilomar Conference of Signals, Systems & Computers, vol. 2, Publ by IEEE, pp. 1651-1655, Proceedings of the 27th Asilomar Conference on Signals, Systems & Computers, Pacific Grove, CA, USA, 11/1/93.

**Deterministic annealing algorithm for entropy-constrained vector quantizer design.** / Rose, Kenneth; Miller, David.

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution

TY - GEN

T1 - Deterministic annealing algorithm for entropy-constrained vector quantizer design

AU - Rose, Kenneth

AU - Miller, David

PY - 1993/12/1

Y1 - 1993/12/1

N2 - In this paper, we develop a deterministic annealing approach for the design of entropy-constrained vector quantizers. Here, the quantization problem is reformulated within a probabilistic setting and the principle of maximum entropy is invoked. The resulting method is connected to statistical physics in two important respects. First, there is annealing in the process controlled by a `temperature' parameter, useful for avoiding local optima of the cost. Second, the number of distinct codevectors grows by a sequence of phase transitions as the temperature is varied. When the temperature is lowered to zero, the method searches for the optimal entropy-constrained vector quantizer. Conditions for bifurcation in the process provide insights into the `right' codebook size for a given problem.

AB - In this paper, we develop a deterministic annealing approach for the design of entropy-constrained vector quantizers. Here, the quantization problem is reformulated within a probabilistic setting and the principle of maximum entropy is invoked. The resulting method is connected to statistical physics in two important respects. First, there is annealing in the process controlled by a `temperature' parameter, useful for avoiding local optima of the cost. Second, the number of distinct codevectors grows by a sequence of phase transitions as the temperature is varied. When the temperature is lowered to zero, the method searches for the optimal entropy-constrained vector quantizer. Conditions for bifurcation in the process provide insights into the `right' codebook size for a given problem.

UR - http://www.scopus.com/inward/record.url?scp=0027874442&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=0027874442&partnerID=8YFLogxK

M3 - Conference contribution

AN - SCOPUS:0027874442

SN - 0818641207

T3 - Conference Record of the Asilomar Conference of Signals, Systems & Computers

SP - 1651

EP - 1655

BT - Conference Record of the Asilomar Conference of Signals, Systems & Computers

PB - Publ by IEEE

ER -