DocumentCode
3256477
Title
Designing max-min propagation neural networks by hyperplane switching
Author
Estévez, Pablo A.
Author_Institution
Dept. of Electr. Eng., Chile Univ., Santiago, Chile
Volume
2
fYear
1995
fDate
29 Nov-1 Dec 1995
Firstpage
596
Abstract
A method for synthesizing max-min propagation neural networks by using genetic algorithms is proposed. These networks are viewed as switching among hyperplanes and the switching configurations are evolved. A distance measure between n-ary strings of variable length is introduced. This metric is used in a niching algorithm to find multiple optima in the space of architectures. Simulation results on the exclusive-OR problem and the interpretation of electrocardiograms are presented
Keywords
electrocardiography; genetic algorithms; medical signal processing; minimax techniques; neural nets; switching; ECG interpretation; architecture space; distance measure; exclusive-OR problem; genetic algorithms; hyperplane switching; max-min propagation neural network synthesis; multiple optima; n-ary strings; niching algorithm; simulation; switching configuration evolution; variable length strings; Biological cells; Electronic mail; Genetic algorithms; Jacobian matrices; Length measurement; Network synthesis; Neural networks; Prototypes; Switches;
fLanguage
English
Publisher
ieee
Conference_Titel
Evolutionary Computation, 1995., IEEE International Conference on
Conference_Location
Perth, WA
Print_ISBN
0-7803-2759-4
Type
conf
DOI
10.1109/ICEC.1995.487451
Filename
487451
Link To Document