Title :
Some qualitative aspects of neural computing circuits
Author_Institution :
Comput. & Autom. Inst. of the Hungarian Acad. of Sci., Budapest, Hungary
Abstract :
The unique and stable computed solutions of a neural circuit are considered. It is shown that under reasonable conditions the row-sum dominance and off-diagonal antitonicity of (-T), the feedback connection matrix, are sufficient conditions. A measure for determining the finite physical complexity is introduced and estimated
Keywords :
feedback; neural nets; stability criteria; artificial intelligence; feedback connection matrix; finite physical complexity; neural computing circuits; off diagonal antitonicity; off-diagonal antitonicity; row-sum dominance; stable computed solutions; Asymptotic stability; Automation; Circuit stability; Digital-analog conversion; Neurofeedback; Nonlinear circuits; Physics computing; State feedback; Sufficient conditions; Voltage control;
Conference_Titel :
Circuits and Systems, 1988., IEEE International Symposium on
Conference_Location :
Espoo
DOI :
10.1109/ISCAS.1988.15034
