Title :
On a continuous-type associative memory
Author :
Surina, Irina I.
Author_Institution :
Kurchatov Inst., Russian Nat. Res. Center, Moscow, Russia
Abstract :
Some considerations on a notion of a associative memory are first presented. A novel model of an associative memory with complex-valued neurons is then presented. The proposed model, called an associative complex-valued neural network (ACNN), and its variations can lead to many useful applications. It has a clear algebraic structure which helps in the analysis of its properties. It is suggested that it might be possible to find an energy function resembling Hopfield´s and that learning without spurious memories might be constructed in this model. Hopfield´s model is embedded into the ACNN and can be examined in this context
Keywords :
Hopfield neural nets; content-addressable storage; learning systems; Hopfield´s model; algebraic structure; associative complex-valued neural network; complex-valued neurons; continuous-type associative memory; energy function; learning; Associative memory; Ducts; Electrical capacitance tomography; Multi-layer neural network; Neural networks; Neurons; Quantum computing;
Conference_Titel :
Neuroinformatics and Neurocomputers, 1992., RNNS/IEEE Symposium on
Conference_Location :
Rostov-on-Don
Print_ISBN :
0-7803-0809-3
DOI :
10.1109/RNNS.1992.268640
