Abstract
Two new algorithms and associated neuron-like network architectures are proposed for solving the eigenvalue problem in real-time. The first approach is based on the solution of a set of nonlinear algebraic equations by employing optimization techniques. The second approach employs a multilayer neural network with linear artificial neurons and it exploits the continuous-time error back-propagation learning algorithm. The second approach enables us to find all the eigenvalues and the associated eigenvectors simultaneously by training the network to match some desired patterns, while the first approach is suitable to find during one run only one particular eigenvalue (e.g. an extreme eigenvalue) and the corresponding eigenvector in realtime. In order to find all eigenpairs the optimization process must be repeated in this case many times for different initial conditions. The performance and convergence behaviour of the proposed neural network architectures are investigated by extensive computer simulations.
Similar content being viewed by others
References
Auchmuty G (1989) Unconstrained variational principles for eigenvalues of real symmetric matrices. SIAM J Math Anal 20:1186–1207
Cichocki A, Unbehauen R (1991a) Switched-Capacitor neural networks for differential optimization. Int J Circuit Theory Appl 19:161–167
Cichocki A, Unbehauen R (1991b) Winner-take-all neural network and its application to minimax optimization problems. Electron Lett 27:2026–2028
Cichocki A, Unbehauen R (1992a) Neural networks for solving systems of linear equations and related problems. IEEE Trans Circuits Syst 39:124–198
Cichocki A, Unbehauen R (1992b) Neural networks for optimization, computing and signal processing, Verlag Teubner-Wiley, Stuttgart
Comon P, Golub GH (1990) Tracking a few extreme singular values and vectors in signal processing. Proc IEEE 78:1327–1343
Evans DJ, Margaritis K (1989) Systolic design for eigenvalue-eigenvector computations using matrix powers. Parallel Comput 14:77–87
Fletcher R (1987) Practical methods of optimization. J. Wiley, New York
Golub GH, Van Loan CF (1989) Matrix computations. Johns Hopkins University Press, Baltimore
Hopfield JJ (1984) Neurons with graded response have collective computational properties like those of two-state neurons. Proc Natl Acad Sci USA 81:3088–3092
Lippman RP (1987) An introduction to computing with neural nets. IEEE ASSP Magz 1987:4–22
Oja E, Karhunen J (1985) On stochastic approximation of the eigenvectors and eigenvalues of the expectation of a random matrix. J Math Anal Appl 106: 69–84
Parlett BN (1980) The symmetric eigenvalue problem. Prentice-Hall, Englewood Cliffs
Rumelhart DE, McCleland JL (eds) (1986) Parallel Distributed Processing. The MIT Press, Cambridge
Samardzija N, Waterland RL (1991) A neural network for computing eigenvectors and eigenvalues. Biol Cybern 65:211–214
Tank DW, Hopfield JJ (1986) Simple neural optimization networks: an A/D converter, signal decision circuit, and a linear programming circuit. IEEE Trans Circuits Syst CAS 33:533–541
Unbehauen R, Cichocki A (1989) MOS switched-capacitor and continuous-time circuits and systems. Springer, Berlin Heidelberg New York
Wang L, Mendel JM (1990) Matrix computations and equation solving using structured networks and training. Proc IEEE Conf. on Decision and Control, Honolulu (USA) pp 1747–1750
Yang X, Sarkar TK, Arvas E (1989) A survey of conjugate gradient algorithms for solution of extreme eigen-problems of a symmetric matrix. IEEE Trans Acoust Speech, Signal Process 37:1550–1555
Author information
Authors and Affiliations
Rights and permissions
About this article
Cite this article
Cichocki, A., Unbehauen, R. Neural networks for computing eigenvalues and eigenvectors. Biol. Cybern. 68, 155–164 (1992). https://doi.org/10.1007/BF00201437
Received:
Accepted:
Issue Date:
DOI: https://doi.org/10.1007/BF00201437