A method to determine the required number of neural-network training repetitions
Authors:
- R.R. RhinehartRRR.R. Rhinehart
- 1 March 1999
- journal article
- research article
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Neural Networks
- Vol. 10 (2) , 427-432
Abstract
Conventional neural-network training algorithms often get stuck in local minima. To find the global optimum, training is conventionally repeated with ten, or so, random starting values for the weights. Here we develop an analytical procedure to determine how many times a neural network needs to be trained, with random starting weights, to ensure that the best of those is within a desirable lower percentile of all possible trainings, with a certain level of confidence. The theoretical developments are validated by experimental results. While applied to neural-network training, the method is generally applicable to nonlinear optimization.Keywords
This publication has 12 references indexed in Scilit:
- Application of neural network control to distillation and an experimental comparison with other advanced controllersISA Transactions, 1999
- Identification and control of anaerobic digesters using adaptive, on-line trained neural networksComputers & Chemical Engineering, 1997
- EXPERIMENTAL COMPARISON OF CONTROL STRATEGIESControl Engineering Practice, 1997
- Critical values for a steady-state identifierJournal of Process Control, 1997
- Automated stopping criteria for neural network training1997
- Predicting sun spots using a layered perceptron neural networkIEEE Transactions on Neural Networks, 1996
- An efficient method for on-line identification of steady stateJournal of Process Control, 1995
- A very simple structure for neural network control of distillationJournal of Process Control, 1995
- Bioprocess optimization and control: Application of hybrid modellingJournal of Biotechnology, 1994
- An Algorithm for Least-Squares Estimation of Nonlinear ParametersJournal of the Society for Industrial and Applied Mathematics, 1963