Singular Value Decomposition Learning on Double Stiefel Manifold
Author:
- 1 June 2003
- journal article
- research article
- Published by World Scientific Pub Co Pte Ltd in International Journal of Neural Systems
- Vol. 13 (03) , 155-170
Abstract
The aim of this paper is to present a unifying view of four SVD-neural-computation techniques found in the scientific literature and to present some theoretical results on their behavior. The considered SVD neural algorithms are shown to arise as Riemannian-gradient flows on double Stiefel manifold and their geometric and dynamical properties are investigated with the help of differential geometry.Keywords
This publication has 21 references indexed in Scilit:
- A theory for learning based on rigid bodies dynamicsIEEE Transactions on Neural Networks, 2002
- Improved RAN sequential prediction using orthogonal techniquesNeurocomputing, 2001
- Image compression using principal component neural networksImage and Vision Computing, 2001
- A Theory for Learning by Weight Flow on Stiefel-Grassman ManifoldNeural Computation, 2001
- Cross-correlation neural network modelsIEEE Transactions on Signal Processing, 1994
- Singular-value decomposition via gradient and self-equivalent flowsLinear Algebra and its Applications, 1992
- Dynamical systems that sort lists, diagonalize matrices, and solve linear programming problemsLinear Algebra and its Applications, 1991
- Design of two-dimensional FIR digital filters by using the singular-value decompositionIEEE Transactions on Circuits and Systems, 1990
- On the finite escape phenomena for matrix Riccati equationsIEEE Transactions on Automatic Control, 1982
- Principal component analysis in linear systems: Controllability, observability, and model reductionIEEE Transactions on Automatic Control, 1981