I looking into optimize the neural network weights. Which neural network model is suitable for prediction based issues. What optimization algorithm gives better result. Can anyone help me with any references and example codes?
Thanks in advance.
Neural networks are general function approximators, so yes, they can. We have used multilayer perceptrons with backpropagation as well as Radial Basis Functions. They both work quite well. Check this paper, for instance
Rivas, V. M., Merelo, J. J., Castillo, P. A., Arenas, M. G., & Castellano, J. G. (2004). Evolving RBF neural networks for time-series forecasting with EvRBF. Information Sciences, 165(3), 207-220.
Yes, Neural Networks can be trained in order to perform forecasting tasks. In order to do so, you need to train the model with samples consisting inputs corresponding to values at t0 and outputs corresponding to values at t1=t0+a, a>0, where a is the interval you want to perform the forecasting ahead of the samples you use.
Neural Networks are extensively used by many researchers in forecasting. Most of the forecasting work done using ANN could be found in Coastal/ Ocean Engineering problems like wave forecasting etc..
Sometimes ANN alone cannot perform well, you need to integrtae ANN algorithm with Optimistaion Algorithms like Swarm Optimisation, Genetic Algorithms etc.. whcih gives better performance than the ANN alone.
you may refer this paper
"Genetic algorithm based support vector machine regression in predicting wave transmission of horizontally interlaced multi-layer moored floating pipe breakwater"
Advances in Engineering Software
Volume 45, Issue 1, March 2012, Pages 203–212
- Badges
- Science topic