The Journal of Korean Institute of Communications and Information Sciences (한국통신학회논문지)

The Korean Institute of Commucations and Information Sciences (한국통신학회)
권호 표지

A New Compensated Criterion in Testing Trained Codebooks

  • Kim, Dong-Sik (School of Electrical and Computer Engineering, Purdue University, West Lafayette, U. S. A. Regular Member)
  • Published : 1999.07.01
Download PDF
⟨ Previous Next ⟩

Abstract

In designing the quantizer of a coding scheme using a training sequence (TS), the training algorithm tries to find a quantizer that minimizes the distortion measured in the TS. In order to evaluate the trained quantizer or compare the coding scheme, we can observe the minimized distortion. However, the minimized distortion is a biased estimate of the minimal distortion for the input distribution. Hence, we could often overestimate a quantizer or make a wrong comparison even if we use a validating sequence. In this paper, by compensating the minimized distortion for the TS, a new estimate is proposed. Compensating term is a function of the training ratio, the TS size to the codebook size. Several numerical results are also introduced for the proposed estimate.

Keywords

References

  1. Clustering Analysis for Applications M.R. Anderberg
  2. Pattern Classification and Scene Analysis R.O. Duda;P.E. Hart
  3. Measure Theory P.R. Halmos
  4. Probability Theory I.(4th. ed) M. Loeve
  5. Principles of Digital Communication and Coding A.J. Viterbi;J.K. Omura
  6. Estimation of Dependences Based on Empirical Data V. Vapnik
  7. Neurcomputing R. Hecht-Nielsen
  8. Vector Quantization and Signal Compression A. Gersho;R.M. Gray
  9. Fundamentals of Speech Recognition L.R. Rabiner;B.-H. Juang
  10. Probability and Measure(3rd ed.) P. Billingsley
  11. Technometrics v.4 On the generation of normal random vectors E.M. Scheuer;D.S. Stoller
  12. IEEE Trans. Inform. Theory v.IT-25 no.4 Asymptotically optimal block quantization A. Gersho
  13. IEEE Trans. Inform. Theory v.IT-28 no.2 Asymptotic quantization error of continuous signals and the quantization dimension P. Zador
  14. IEEE Trans. Inform. Theory v.IT-28 no.2 Voronoi regions of lattices, second moments of polytopes, and quantization J.H. Conway;N.J.A. Sloane
  15. IEEE Trans. Inform. Theory v.IT-28 no.2 Multidimensional asymptotic quantization theory with rth power distortion measures J.A. Bucklew;G.L. Wise
  16. IEEE Trans. Inform. Theory v.IT-28 no.6 On the existence of optimal quantizers E.F. Abaya;G.L. Wise
  17. SIAM J. Alg. Discrete Methods v.4 no.1 The optimal lattice quantizer in three dimensions E.S. Barnes;N.J.A. Sloane
  18. SIAM J. Alg. Discrete Methods v.5 no.3 On the voronoi regions of certain lattices J.H. Conway;N.J.A. Sloane
  19. SIAM J. Appl. Math. v.44 no.1 Convergence of vector quantizers with applications to optimal quantization E.A. Abaya;G.L. Wise
  20. IEEE Trans. Inform. Theory v.IT-31 no.1 A lower bound on the average error of vector quantizers J.H. Conway;N.J.A. Sloane
  21. IEEE Trans. Inform. Theory v.IT-33 Robust source coding of weakly compact classes D.L. Neuhoff;R. Garcia-Munoz
  22. Proc. IEEE v.78 no.9 The self-organizing map T. Kohonen
  23. IEEE Trans. Commun. v.COM-39 no.4 Image vector quantizer based on a classification in the DCT domain D.S. Kim;S.U. Lee
  24. Ph.D. dissertation. Dep. of Control and Instrumentation, Seoul National Univ. Performance of vector quantizers and its relation to training-set size D.S. Kim
  25. IEEE Trans. Image Processing v.IP-6 no.3 On testing trained vector quantizer codebooks D.S. Kim;T. Kim;S.U. Lee
  26. Tech. Rep. TR-ECE 98-6 Bounds for the trained vector quantizer distortion measured in training data D.S. Kim;M.R. Bell
  27. IEEE Trans. Inform. Theory Quantization basedd on a novel sample-adaptive product quantizer (SAPQ) D.S. Kim;N.B. Shroff