Application of SVR with improved ant colony optimization algorithms in exchange rate forecasting

作者：Hung Wei Mou^*; Hong Wei Chiang

来源：Control and Cybernetics, 2009, 38(3): 863-891.

• support vector regression (SVR)
• continuous ant colony optimization algorithms (CACO)
• exchange rates
• financial forecasting

Traditional time series forecasting models, like ARIMA and regression models, can hardly capture nonlinear patterns. Support vector regression (SVR), a novel neural network technique, has been successfully used to solve nonlinear regression and time series problems. The SVR model applies the structural risk minimization principle to minimize the upper bound of the generalization error, instead of minimizing the training error, employed by most conventional neural network models. Thus, parameter determination for an SVR model is appropriate for achieving high forecasting accuracy. Several evolutionary algorithms, such as genetic algorithms and simulated annealing algorithms have been used in parameter selection, but these algorithms often suffer from the possibility of being trapped in local optimum. This study used an improved ant colony optimization algorithm in an SVR model, called SVRCACO, for selecting suitable parameters, with encouraging local search in areas where forecasting accuracy improvement continues to be made, then, autocatalytically converge to promising regions. Numerical examples of exchange rate forecasting from the existing literature are employed to assess the performance of the proposed model. Experimental results show that the proposed model outperforms other approaches from the literature.

• 出版日期2009