Title :
Reinforced Two-Step-Ahead Weight Adjustment Technique for Online Training of Recurrent Neural Networks
Author :
Li-Chiu Chang ; Pin-An Chen ; Chang, Fi-John
Author_Institution :
Dept. of Water Resources & Environ. Eng., Tamkang Univ., Taipei, Taiwan
Abstract :
A reliable forecast of future events possesses great value. The main purpose of this paper is to propose an innovative learning technique for reinforcing the accuracy of two-step-ahead (2SA) forecasts. The real-time recurrent learning (RTRL) algorithm for recurrent neural networks (RNNs) can effectively model the dynamics of complex processes and has been used successfully in one-step-ahead forecasts for various time series. A reinforced RTRL algorithm for 2SA forecasts using RNNs is proposed in this paper, and its performance is investigated by two famous benchmark time series and a streamflow during flood events in Taiwan. Results demonstrate that the proposed reinforced 2SA RTRL algorithm for RNNs can adequately forecast the benchmark (theoretical) time series, significantly improve the accuracy of flood forecasts, and effectively reduce time-lag effects.
Keywords :
learning (artificial intelligence); recurrent neural nets; time series; 2SA; RNN; RTRL; flood forecasts; online recurrent neural networks training; real-time recurrent learning algorithm; reinforced two-step-ahead weight adjustment technique; time series; time-lag effects; two-step-ahead forecasts; Algorithm design and analysis; Predictive models; Recurrent neural networks; Time series analysis; Training; Real-time recurrent learning (RTRL) algorithm; recurrent neural network (RNN); streamflow forecast; time series forecast;
Journal_Title :
Neural Networks and Learning Systems, IEEE Transactions on
DOI :
10.1109/TNNLS.2012.2200695
