In this paper, we present a modified FIR (Finite Impulse Response) network model for improving the capability of time series prediction system. The model has interval arithmetic capability as well as the time series prediction capability of the Finite Impulse Response (FIR) network. The proposed model exhibits some advantageous features, as follows. Since the interval values can be generated as input features for the neural network by data segmentation and grouping, the amount of data and computation for the learning stage can be reduced. The weather forecast system based on the model can generate the output values in the form of interval representation, and can avoid the over-training effect that is caused by unbalanced learning data. From the experimental results of the forecast of monthly regional precipitation in Korea, the usefulness of the proposed model is evaluated.
[1]
Eric A. Wan,et al.
Time series prediction by using a connectionist network with internal delay lines
,
1993
.
[2]
Fionn Murtagh,et al.
Dynamical recurrent neural networks -- towards environmental time series prediction
,
1995,
Int. J. Neural Syst..
[3]
Geoffrey E. Hinton,et al.
Phoneme recognition using time-delay neural networks
,
1989,
IEEE Trans. Acoust. Speech Signal Process..
[4]
K. Nakayama,et al.
Interval arithmetic backpropagation
,
1993,
Proceedings of 1993 International Conference on Neural Networks (IJCNN-93-Nagoya, Japan).
[5]
Emanuel Marom,et al.
Efficient Training of Recurrent Neural Network with Time Delays
,
1997,
Neural Networks.