Journal of IKEEE (전기전자학회논문지)

Institute of Korean Electrical and Electronics Engineers (한국전기전자학회)
권호 표지

Neural Learning Algorithms for Independent Component Analysis

  • Choi, Seung-Jin (School of Electrical and Electronics Engineering, Chungbuk National University)
  • 최승진 (충북대학교 전기전자공학부)
  • Published : 1998.08.01
Download PDF
⟨ Previous Next ⟩

Abstract

Independent Component analysis (ICA) is a new statistical method for extracting statistically independent components from their linear instantaneous mixtures which are generated by an unknown linear generative model. The recognition model is learned in unsupervised manner so that the recovered signals by the recognition model become the possibly scaled estimates of original source signals. This paper addresses the neural learning approach to ICA. As recognition models a linear feedforward network and a linear feedback network are considered. Associated learning algorithms for both networks are derived from maximum likelihood and information-theoretic approaches, using natural Riemannian gradient [1]. Theoretical results are confirmed by extensive computer simulations.

Keywords

References

  1. Neural Computation v.10 Natural gradient works efficiently in learning Amari, S.
  2. Signal Processing v.36 Independent component analysis, a new concept? Comon, P.
  3. Signal Processing v.24 Blind separation of sources, part I: An adaptive algorithm based on neuromimetic architecture Jutten, C.;Herault, J.
  4. International Joint Conference on Neural Networks A new on-line adaptive learning algorithm for blind separation of source signals Cichocki, A.;Unbehauen, R.;Moszczynski, L.;Rummert, E.
  5. Neural Computation v.7 An information maximization approach to blind separation and blind deconvolution Bell, A.;Sejnowski, T.
  6. European Signal Processing Conference Separation of mixture of independent sources through a maximum likelihood approach Pham, D.T.;Garrat, P.;Jutten, C.
  7. Maximum likelihood and covariant algorithms for independent component analysis;Draft 3.7 MacKay, D.J.C.
  8. IEEE Trans. Signal Processing v.44 Equivariant adaptive source separation Cardoso, J.F.;Laheld, B.H.
  9. Advances in Neural Information Processing Systems v.8 A new learning algorithm for blind signal separation Amari, S.;Cichocki, A.;Yang, H.H.;Touretzky, D.S.(ed.);Mozer, M.C.(ed.);Hasselmo, M.E.(ed.)
  10. Neural Processing Letters v.7 A spurious equlibria-free learning algorithm for the blind separation of non-zero skewness signals Choi, S.;Liu, R.;Cichocki, A.
  11. IEEE Signal Processing Workshop on Higher-order Statistics A linear feedforward neural network with lateral feedback connections Choi, S.;Cichocki, A.
  12. Elements of Informacion Theory Cover, T.M.;Thomas, J.A.
  13. IEEE Signal Processing Letters v.4 Infomax and maximum likelihood for source separation Cardoso, J.F.
  14. Network: Computation in Neural Systems v.5 Nonlinear neurons in the low-noise limit: A factorial code maximize information transfer Nadal, J.P.;Parga, N.
  15. Adaptive blind signal separation for multiuser communications: An Information-theoretic approach Choi, S.