Title :
An adaptive quasi-Newton algorithm for eigensubspace estimation
Author :
Kang, Zhengjiu ; Chatterjee, Chanchal ; Roychowdhury, Vwani P.
Author_Institution :
Dept. of Electr. Eng., California Univ., Los Angeles, CA, USA
fDate :
12/1/2000 12:00:00 AM
Abstract :
We derive and discuss a new adaptive quasi-Newton eigen-estimation algorithm and compare it with the RLS-type adaptive algorithms and the quasi-Newton algorithm proposed by Mathew et al. (1995) through experiments with stationary and nonstationary data.
Keywords :
Newton method; adaptive estimation; adaptive signal processing; convergence of numerical methods; eigenvalues and eigenfunctions; principal component analysis; RLS-type adaptive algorithms; adaptive quasi-Newton eigen-estimation algorithm; convergence analysis; eigensubspace estimation; experiments; nonstationary data; principal component analysis; stationary data; Adaptive algorithm; Algorithm design and analysis; Approximation algorithms; Cost function; Covariance matrix; Eigenvalues and eigenfunctions; Neural networks; Optimization methods; Principal component analysis; Symmetric matrices;
Journal_Title :
Signal Processing, IEEE Transactions on
