The KIPS Transactions:PartB (정보처리학회논문지B)

Korea Information Processing Society (한국정보처리학회)
권호 표지
DOI QR코드

DOI QR Code

An Algorithm of Score Function Generation using Convolution-FFT in Independent Component Analysis

독립성분분석에서 Convolution-FFT을 이용한 효율적인 점수함수의 생성 알고리즘

  • 김웅명 (경희대학교 컴퓨터공학과) ;
  • 이현수 (경희대학교 컴퓨터공학과)
  • Published : 2006.02.01
Download PDF
⟨ Previous Next ⟩

Abstract

In this study, we propose this new algorithm that generates score function in ICA(Independent Component Analysis) using entropy theory. To generate score function, estimation of probability density function about original signals are certainly necessary and density function should be differentiated. Therefore, we used kernel density estimation method in order to derive differential equation of score function by original signal. After changing formula to convolution form to increase speed of density estimation, we used FFT algorithm that can calculate convolution faster. Proposed score function generation method reduces the errors, it is density difference of recovered signals and originals signals. In the result of computer simulation, we estimate density function more similar to original signals compared with Extended Infomax and Fixed Point ICA in blind source separation problem and get improved performance at the SNR(Signal to Noise Ratio) between recovered signals and original signal.

본 연구에서는 엔트로피를 이용한 독립성분분석(ICA : Independent Component Analysis)에서 점수함수(score function)를 생성하는 알고리즘을 제안한다. 점수함수를 생성하기 위해서 원 신호(original signals)에 대한 확률밀도함수의 추정이 반드시 필요하고 밀도함수가 미분 가능해야 한다. 따라서 원 신호에 따른 적응적인 점수 함수를 유도할 수 있도록 커널 기반의 밀도추정(kernel density estimation)방법을 사용하였으며, 보다 빠른 밀도 추정 계산을 위해서 식의 형태를 컨볼루션(convolution) 변환 한 후, 컨볼루션을 빠르게 계산할 수 있는 FFT(Fast Fourier Transform) 알고리즘을 이용하였다. 제안한 점수함수 생성 방법은 원 신호에 확률밀도분포와 추정된 신호의 확률밀도 분포의 오차를 줄이는 역할을 한다 실험 결과, 암묵신호분리(blind source separation)문제에서 기존의 Extended Infomax 알고리즘과 Fixed Point ICA 보다 원 신호와 유사한 밀도함수를 추정하였고, 분리된 신호의 신호대잡음비등(SNR)에 있어서 향상된 성능을 얻을 수 있었다.

Keywords

References

  1. P. Comon, 'Independent Component Analysis, A New Concept?,' Signal Processing, Vol.36, No.3, pp.287-314, 1994 https://doi.org/10.1016/0165-1684(94)90029-9
  2. A. Hyv？rinen, J. Karhunen, Erkki Oja, 'Independent Component Analysis', Wiley interscience, 2001
  3. S.-i. Amari, A. Cichocki, and H. H. Yang, 'A New Learning Algorithm for Blind Source Separation,' Advances in Neural Information Processing Systems. Cambridge, MA: MIT Press, Vol.8, pp.757-763, 1996
  4. A. J. Bell and T. Sejnowski, 'An Information-Maximization Approach to Blind Separation and Blind Deconvolution,' Neural Computation, Vol.7, No.6, pp.1129-1159, 1995 https://doi.org/10.1162/neco.1995.7.6.1129
  5. A. J. Bell and T. J. Sejnowski, 'The 'Independent Components' of Natural Scenes are Edge Filters,' Vision Research, Vol.37, pp.3327-3338, 1997 https://doi.org/10.1016/S0042-6989(97)00121-1
  6. J.-F. Cardoso, 'Blind Signal Separation: Statistical Principles,' Proc. IEEE. Special Issue on Blind Identification and Estimation, Vol.9, pp.2009-2025, Oct. 1998 https://doi.org/10.1109/5.720250
  7. J.-F. Cardoso, 'Higher-Order Contrasts for Independent Component Analysis,' Neural Computation, Vol.11, No.1, pp.157-192, Jan., 1999 https://doi.org/10.1162/089976699300016863
  8. M. Girolami and C. Fyfe, 'An Extended Exploratory Projection Pursuit Network with Linear and Nonlinear anti-Hebbian Lateral Connections Applied to the Cocktail Party Problem,' Neural Networks, Vol.10, No.9, pp.1607-1618, 1997 https://doi.org/10.1016/S0893-6080(97)00090-7
  9. D. Obradovic and G. Deco, 'Information Maximization and Independent Component Analysis: Is There a Difference?,' Neural Computation, Vol.10, pp.2085-2101, 1998 https://doi.org/10.1162/089976698300016972
  10. T.-W. Lee, M. Girolami, and T. J. Sejnowski, 'Independent Component Analysis using an Extended Infomax Algorithm for Mixed sub-Gaussian and super-Gaussian Sources,' Neural Computation, Vol.11, No.2, pp.417-441, 1999 https://doi.org/10.1162/089976699300016719
  11. S. Choi, A. Cichocki, and S. Amari, 'Flexible Independent Component Analysis', Journal of VLSI Signal Processing, Vol.26, No.2, pp.25-38, August, 2000 https://doi.org/10.1023/A:1008135131269
  12. S. Makeig, T.-P. Jung, A. J Bell, D. Ghahremani, and T. J. Sejnowski, 'Blind Separation of Event-related Brain Responses into Independent Components,' in Proc. National Academic. Sciences USA, Vol.94, pp.979-984, 1997 https://doi.org/10.1073/pnas.94.20.10979
  13. T.-P. Jung, C. Humphries, T.-W. Lee, M. J. McKeown, V. Iragui, S. Makeig, and T. J. Sejnowski, 'Removing electroencephalographic artifacts by blind source separation', Psychophysiology., Vol.37, pp.163-178, 2000 https://doi.org/10.1017/S0048577200980259
  14. B. W. Silverman, 'Density Estimation for Statistics and Data Analysis' New York: Chapman and Hall, 1985
  15. N. Vlassis and Y. Motomura, 'Efficient source adaptivity in independent component analysis,' IEEE Trans. Neural Networks, Vol.12, pp.559-566, May.200l https://doi.org/10.1109/72.925558
  16. S. Fiori and P. Bucciarelli, 'Probability Density Estimation Using Adaptive Activation Function Neurons', Neural Processing Letters, Vol.13, No.1, pp.31-42, Feb., 2001 https://doi.org/10.1023/A:1009635129159
  17. Woong Myung Kim and Hyon Soo Lee, 'An Efficient Score Function Generation Algorithm with Information Maximization', Advances in Natural Computation. Lecture Note on Computer Science, Vol.3610, pp.760-768, 2005 https://doi.org/10.1007/11539087_100