Title :
Dual systems for minor and principal component computation
Author :
Hasan, Mohammed A.
Author_Institution :
Dept. of Electr. Sz Comput. Eng., Univ. of Minnesota, Duluth, MN
fDate :
March 31 2008-April 4 2008
Abstract :
Converting principal component dynamical system to a minor component dynamical system and vice versa sometimes leads to unstable systems. In this paper, classes of globally stable dynamical systems that can be converted between PCA and MCA systems by merely switching the signs of some terms of a given system are developed. These systems are shown to be applicable to symmetric and nonsymmetric matrices. These systems are then modified to be asymptotically stable by adding a penalty term. The proposed systems may apply to both the standard and the generalized eigenvalue problems. Lyapunov stability theory and LaSalle invariance principle are used to derive invariant sets for these systems.
Keywords :
Lyapunov matrix equations; asymptotic stability; duality (mathematics); eigenvalues and eigenfunctions; principal component analysis; LaSalle invariance principle; Lyapunov stability theory; asymptotical stability; dual systems; generalized eigenvalue problems; global stability; minor component analysis; minor component dynamical system; nonsymmetric matrice; principal component analysis; principal component dynamical system; symmetric matrices; Convergence; Eigenvalues and eigenfunctions; Equations; Lyapunov method; Matrix converters; Nonlinear systems; Principal component analysis; Symmetric matrices; Lyapunov stability; Oja’s learning rule; Principal components; Rayleigh quotient; dual-purpose MCA/PCA systems; generalized eigenvalue problem; global convergence; minor components;
Conference_Titel :
Acoustics, Speech and Signal Processing, 2008. ICASSP 2008. IEEE International Conference on
Conference_Location :
Las Vegas, NV
Print_ISBN :
978-1-4244-1483-3
Electronic_ISBN :
1520-6149
DOI :
10.1109/ICASSP.2008.4518006