Title :
Recursive procedure based on virtual state method for estimating connection value of nonlinear circuit associative network
Author :
Suetsugu, Tadashi ; Tanaka, Mamoru ; Mori, Shinsaku
Author_Institution :
Dept. of Electr. Eng., Keio Univ., Yokohama, Japan
Abstract :
The authors describe a novel associative network model of neural networks, and a novel algorithm for estimating the singular connecting matrix using the virtual state method in network learning. By this method the nonsingularized coefficient matrix can be estimated recursively by the escalator method. Therefore, learning the (k+1)th pattern can be done just by modifying the inverse matrix of the kth pattern slightly
Keywords :
learning systems; matrix algebra; neural nets; nonlinear network analysis; connection value; escalator method; network learning; network model; neural networks; nonlinear circuit associative network; nonsingularized coefficient matrix; recursive procedure; singular connecting matrix; virtual state method; Circuit topology; Equations; Least squares approximation; Least squares methods; Piecewise linear approximation; Recursive estimation; Resistors; Sparse matrices; State estimation; Vectors;
Conference_Titel :
Circuits and Systems, 1991., Proceedings of the 34th Midwest Symposium on
Conference_Location :
Monterey, CA
Print_ISBN :
0-7803-0620-1
DOI :
10.1109/MWSCAS.1991.252062
