A Meta Neural Network Polling System for the RPROP Learning Rule
This paper proposes an application independent method of automating learning rule parameter selection using a group of supervisor neural networks, known as meta neural networks, to alter the value of a learning rule parameter during training. Each meta neural network is trained using data generated by observing the training of a neural network and recording the effects of the selection of various parameter values. A group of meta neural networks is then polled to obtain a parameter value for a learning rule. Experiments are undertaken to see how this method performs by using it to adapt a global parameter of RPROP.
KeywordsLearning Rule Heart Problem Rule Parameter Supervisor Neural Network Thyroid Problem
Unable to display preview. Download preview PDF.
- C. McCormack. A study of the adaptation of learning rule parameters using a meta neural network. In 13th European Meeting on Systems and Cybernetic Research, volume 2, pages 1043–1048, 1996.Google Scholar
- C. McCormack. Using a meta neural network for RPROP parameter adaptation. In Proc. European Symposium on Artificial Neural Networks, pages 7–12, 1996.Google Scholar
- I. Pitas. Parallel Algorithms for Digital Image Processing, Vision and Neural Networks. John Wiley and Sons, Chichester, 1993.Google Scholar
- L. Prechelt. PROBEN1: A set of neural network benchmarking rules. Technical Report 21/94, Dept. of Informatics, University of Karlsruhe, Germany, 1994. ftp://ftp.ira.uka.de/pub/neuron/probenl.tar.gz.Google Scholar