Title :
An active associative memory neural network model
Author_Institution :
Inst. of Autom., Acad. Sinica, Beijing, China
fDate :
27 Jun-2 Jul 1994
Abstract :
Inspiring of the facts that the information coding can improve the capacity of associative memory and the active mechanism in the brain, an active associative memory model is proposed in this paper. The model consists of two neural networks which can produce endogenous patterns in the learning phase. The model gives good performance because the endogenous patterns are independent from the input patterns and can be approximate orthogonal. Preliminary analysis and computer simulations showed that the model has much higher capacity then the conventional associative memory models, specially in the case in which the input patterns are highly correlative. Finally, the relationships between the model and bidirectional associative memory and hologram are discussed
Keywords :
content-addressable storage; learning (artificial intelligence); neural nets; active associative memory; bidirectional associative memory; endogenous patterns; hologram; learning phase; neural network model; Associative memory; Biological neural networks; Brain modeling; Computer simulation; Neural network hardware; Neural networks; Neurons; Pattern analysis; Performance analysis; Plastics;
Conference_Titel :
Neural Networks, 1994. IEEE World Congress on Computational Intelligence., 1994 IEEE International Conference on
Conference_Location :
Orlando, FL
Print_ISBN :
0-7803-1901-X
DOI :
10.1109/ICNN.1994.374350