Title :
Trainability analysis of a Hopfield neural network based on interacting basins of attraction
Author :
Yüksel, Ömür ; Bas, Kivanç
Author_Institution :
Dept. of Electr. & Electron. Eng., Middle East Tech. Univ., Ankara, Turkey
Abstract :
A Hopfield neural network is analysed in order to determine the types of its equilibrium points. Equilibrium points are considered as the intersection of two curves and the slope of these curves are utilised as criteria on the type of the singularities
Keywords :
Hopfield neural nets; learning (artificial intelligence); Hopfield neural network; equilibrium points; interacting basins of attraction; singularities; trainability analysis; Artificial neural networks; Circuits; Eigenvalues and eigenfunctions; Electronic mail; Hopfield neural networks; Jacobian matrices; Neural networks; Nonlinear equations; Shape; State-space methods;
Conference_Titel :
Electrotechnical Conference, 1994. Proceedings., 7th Mediterranean
Conference_Location :
Antalya
Print_ISBN :
0-7803-1772-6
DOI :
10.1109/MELCON.1994.380919
