Title :
Synaptic weight noise during multilayer perceptron training: fault tolerance and training improvements
Author :
Murray, Alan F. ; Edwards, Peter J.
Author_Institution :
Dept. of Electr. Eng., Edinburgh Univ., UK
fDate :
7/1/1993 12:00:00 AM
Abstract :
The authors develop a mathematical model of the effects of synaptic arithmetic noise in multilayer perceptron training. Predictions are made regarding enhanced fault-tolerance and generalization ability and improved learning trajectory. These predictions are subsequently verified by simulation. The results are perfectly general and have profound implications for the accuracy requirements in multilayer perceptron (MLP) training, particularly in the analog domain
Keywords :
fault tolerant computing; feedforward neural nets; generalisation (artificial intelligence); learning (artificial intelligence); analog domain; fault tolerance; generalization; learning; mathematical model; multilayer perceptron training; neural nets; synaptic weight noise; Cognition; Electrons; Fault tolerance; Feedforward neural networks; Learning systems; Microstructure; Multilayer perceptrons; Neural networks; Optimized production technology; Spectroscopy;
Journal_Title :
Neural Networks, IEEE Transactions on
