Title :
The neural network self-healing process by using a reconstructed sample space
Author :
Hodges, Russel E. ; Wu, Chwan-Hwa
Author_Institution :
Dept. of Electr. Eng., Auburn Univ., AL, USA
Abstract :
The inherent ability of neural networks to recover information when neurons in the network are damaged is discussed. This self-healing property is shown to exist in networks used for pattern recognition. The self-organizing feature map (SOFM) is the network used to study this topic. The SOFM has an efficient data compression technique that allows signal processing techniques to be used in recovering lost information from destroyed nodes
Keywords :
data compression; neural nets; pattern recognition; data compression technique; destroyed nodes; neural network; pattern recognition; reconstructed sample space; self-healing process; self-organizing feature map; signal processing techniques; Artificial neural networks; Biological system modeling; Biomedical signal processing; Data compression; Density functional theory; Density measurement; Fault tolerant systems; Neural networks; Neurons; Pattern recognition;
Conference_Titel :
Circuits and Systems, 1990., IEEE International Symposium on
Conference_Location :
New Orleans, LA
DOI :
10.1109/ISCAS.1990.111972
