A comparison between neural network and conventional vector quantization codebook algorithms

Kohonen's (1988) unsupervised learning algorithm is successfully applied to the codebook generation problem. The algorithm has shown to provide a codebook that rivals the performance of the codebooks obtained using the conventional Linde-Buzo-Gray algorithm, while requiring a minimum amount of processing. The unsupervised learning algorithm provides the ability to adapt to changing inputs, something that is not possible with the standard algorithm. These features make Kohonen's unsupervised learning algorithm an attractive alternative to the conventional vector quantization codebook generation technique.<<ETX>>

[1]  Teuvo Kohonen,et al.  Self-Organization and Associative Memory, Second Edition , 1988, Springer Series in Information Sciences.

[2]  R. Gray,et al.  Vector quantization , 1984, IEEE ASSP Magazine.

[3]  R. G. Leonard,et al.  A database for speaker-independent digit recognition , 1984, ICASSP.

[4]  J. MacQueen.  Some methods for classification and analysis of multivariate observations , 1967 .

[5]  J. Vaisey,et al.  Simulated annealing and codebook design , 1988, ICASSP-88., International Conference on Acoustics, Speech, and Signal Processing.

[6]  Robert M. Gray,et al.  An Algorithm for Vector Quantizer Design , 1980, IEEE Trans. Commun..

[7]  Teuvo Kohonen,et al.  The 'neural' phonetic typewriter , 1988, Computer.

[8]  Teuvo Kohonen,et al.  An introduction to neural computing , 1988, Neural Networks.

[9]  S. P. Lloyd,et al.  Least squares quantization in PCM , 1982, IEEE Trans. Inf. Theory.