Oja's algorithm for graph clustering, Markov spectral decomposition, and risk sensitive control
DSpace at IIT Bombay
View Archive InfoField | Value | |
Title |
Oja's algorithm for graph clustering, Markov spectral decomposition, and risk sensitive control
|
|
Creator |
BORKAR, V
MEYN, SP |
|
Subject |
Graph algorithms
Oja's algorithm Stochastic approximation Markov chains Spectral theory of Markov chains Multiplicative ergodic theory Risk sensitive control REVERSIBLE DIFFUSION-PROCESSES STOCHASTIC-APPROXIMATION LEARNING ALGORITHM METASTABILITY CONVERGENCE COST ASYMPTOTICS EIGENVALUES CHAINS |
|
Description |
Given a positive definite matrix M and an integer N-m >= 1, Oja's subspace algorithm will provide convergent estimates of the first N-m eigenvalues of M along with the corresponding eigenvectors. It is a common approach to principal component analysis. This paper introduces a normalized stochastic-approximation implementation of Oja's subspace algorithm, as well as new applications to the spectral decomposition of a reversible Markov chain. Recall that this means that the stationary distribution satisfies the detailed balance equations (Meyn & Tweedie, 2009). Equivalently, the statistics of the process in steady state do not change when time is reversed. Stability and convergence of Oja's algorithm are established under conditions far milder than that assumed in previous work. Applications to graph clustering, Markov spectral decomposition, and multiplicative ergodic theory are surveyed, along with numerical results. (C) 2012 Elsevier Ltd. All rights reserved.
|
|
Publisher |
PERGAMON-ELSEVIER SCIENCE LTD
|
|
Date |
2014-10-17T04:24:56Z
2014-10-17T04:24:56Z 2012 |
|
Type |
Article
|
|
Identifier |
AUTOMATICA, 48(10)2512-2519
http://dx.doi.org/10.1016/j.automatica.2012.05.016 http://dspace.library.iitb.ac.in/jspui/handle/100/15942 |
|
Language |
en
|
|