Title :
Inductive Inference of Invariant Subspaces
Author_Institution :
Department of Electrical Engineering, University of Notre Dame, Notre Dame, Indiana 46556
Abstract :
This paper shows that inductive inference protocols can learn invariant linear subspaces, used in the stabilization of variable structure systems, after a finite number of failed oracle queries. It is further shown that this convergence bound scales in a polynomial manner with the system´s state space dimension.
Keywords :
Convergence; Eigenvalues and eigenfunctions; Equations; Inference algorithms; Iterative algorithms; Machine learning algorithms; Protocols; Symmetric matrices; Variable structure systems; Vectors;
Conference_Titel :
American Control Conference, 1993
Conference_Location :
San Francisco, CA, USA
Print_ISBN :
0-7803-0860-3
