Title :
Physical time-series prediction using second-order pipelined recurrent neural network
Author_Institution :
Sch. of Comput. & Math. Sci., Liverpool John Moores Univ., UK
Abstract :
This paper presents a novel type of higher-order pipelined recurrent neural network called the second-order pipelined recurrent neural network. The aim of the network is to improve the performance of the pipelined recurrent neural network by accommodating second order terms in the inputs. The network is tested for the prediction of non-linear and non-stationary signals. Two physical time-series, which are the mean value of the AE index and the sunspot signals are used in the simulation. The simulation results showed an average improvement in the signal to noise ratio, of 6.09 dB when compared to the pipelined recurrent neural networks.
Keywords :
learning (artificial intelligence); mathematics computing; neural net architecture; recurrent neural nets; signal processing; time series; AE index; learning algorithm; nonlinear signals; nonstationary signals; performance; physical time-series prediction; second order terms; second-order pipelined recurrent neural network; signal to noise ratio; simulation; sunspot signals; Autoregressive processes; Economic forecasting; Environmental economics; Modeling; Neural networks; Power generation economics; Predictive models; Recurrent neural networks; Signal to noise ratio; Testing;
Conference_Titel :
Artificial Intelligence Systems, 2002. (ICAIS 2002). 2002 IEEE International Conference on
Print_ISBN :
0-7695-1733-1
DOI :
10.1109/ICAIS.2002.1048091