Title :
Memory search using tunnelling effect [associative memory]
Author_Institution :
Fac. of Electron. & Telecommun., Tech. Univ., Iasi, Romania
fDate :
5/13/1999 12:00:00 AM
Abstract :
A special associative memory is designed based on the storage of the desired patterns as stable states of a gradient-type dynamic system. They correspond to deep local minima of the associated energy function. The tunnelling effect provides a means of dynamically evolving the state of the system from one minimum to a different, lower energy state
Keywords :
analogue storage; content-addressable storage; recurrent neural nets; tunnelling; associative memory; deep local minima; energy function; gradient-type analogue recurrent network; gradient-type dynamic system; stable states; tunnelling effect;
Journal_Title :
Electronics Letters
DOI :
10.1049/el:19990544
