Title :
Circular neural network: thermodynamics and phase transitions
Author_Institution :
Inst. of Phys. & Technol., Moscow, Russia
Abstract :
Thermodynamics of the circular neural network is studied using mean-field equation for the overlap function. Structure and stability of low-lying states of the spectrum are analysed, with dynamical phase transitions in the system being discussed
Keywords :
Hopfield neural nets; free energy; phase transformations; spectral analysis; stability; Hopfield model; circular neural network; free energy functional; low-lying states; mean-field equation; overlap function; phase transitions; spectral analysis; stability; thermodynamics; Electronic mail; Integral equations; Neural networks; Neurons; Physics; Predictive models; Process design; Stability analysis; Symmetric matrices; Thermodynamics;
Conference_Titel :
Neuroinformatics and Neurocomputers, 1995., Second International Symposium on
Conference_Location :
Rostov on Don
Print_ISBN :
0-7803-2512-5
DOI :
10.1109/ISNINC.1995.480867
