Optimal Wavelet Packets Decomposition Based on a Rate-Distortion Optimality Criterion

Author

Silva, Jorge ; Narayanan, Shrikanth

Author_Institution

Viterbi Sch. of Eng., Univ. of Southern California, CA

Volume

3

fYear

2007

fDate

15-20 April 2007

Abstract

We address the problem of optimal decomposition of wavelet packets (WPs) for pattern recognition based on the minimum probability of error signal representation (MPE-SR) principle. The problem is formulated as a complexity regularized optimization, where the tree-indexed structure of the WP family is used to reduce it to a type of minimum cost tree pruning problem used in regression and classification trees (CART). MPE-SR solutions are obtained for a frame level phone recognition task showing promising performance results.

Keywords

computational complexity; pattern recognition; rate distortion theory; signal classification; signal representation; trees (mathematics); wavelet transforms; classification trees; complexity regularized optimization; frame level phone recognition task; minimum cost tree pruning problem; minimum probability of error signal representation; optimal wavelet packets decomposition; pattern recognition; rate-distortion optimality criterion; tree-indexed structure; Binary trees; Classification tree analysis; Filter bank; Frequency; Pattern recognition; Rate-distortion; Signal representations; Tree graphs; Wavelet analysis; Wavelet packets; Signal representation for classification; Wavelet packets; basis selection; complexity regularization; minimum cost tree pruning;

fLanguage

English

Publisher

ieee

Conference_Titel

Acoustics, Speech and Signal Processing, 2007. ICASSP 2007. IEEE International Conference on

Conference_Location

Honolulu, HI

ISSN

1520-6149

Print_ISBN

1-4244-0727-3

Electronic_ISBN

1520-6149

Type

conf

DOI

10.1109/ICASSP.2007.366805

Filename

4217835