Title :
Associative memory on the basis of high correlation
Author :
Sun, Jifeng ; Nagata, Tadashi
Author_Institution :
Dept. of Electr. Eng., Kyushu Univ., Fukuoka, Japan
Abstract :
The performance of a high-correlation neural network (HCN) as an associative memory is analyzed. The pattern orthogonalization extension by high correlation of the original patterns can improve the ratio of the storage capacity to the total number of connections. The HCN can not only be used in autoassociative memories, but it can also have good performance heteroassociative memories. Simulation results show the effectiveness of HCN
Keywords :
content-addressable storage; neural nets; HCN; associative memory; high correlation; high-correlation neural network; Associative memory; Capacity planning; Hopfield neural networks; Neural networks; Neurofeedback; Neurons; Sun;
Conference_Titel :
Neural Networks, 1992. IJCNN., International Joint Conference on
Conference_Location :
Baltimore, MD
Print_ISBN :
0-7803-0559-0
DOI :
10.1109/IJCNN.1992.287166
