Title :
A robust technique for exchange rate prediction using Wilcoxon Norm
Author :
Mohapatra, U.M. ; Majhi, Banshidhar ; Rout, Minakhi
Author_Institution :
Dept. of CA, Siksha O Anusandhan Univ., Bhubaneswar, India
Abstract :
In this paper, the rank based Wilcoxon norm is used with the neural network to predict exchange rate values when the training data is corrupted with 10%-40% of outliers. The multilayer neural network(MLANN) with Euclidean norm does not seem robust in the presence of outliers. In the contrary, Wilcoxon Norm approach has been proved robust against outliers. Simulation results show improved prediction over MLANN for all exchange rates with different months ahead prediction.
Keywords :
exchange rates; neural nets; Euclidean norm; MLANN; exchange rate prediction; multilayer neural network; rank based Wilcoxon norm approach; robust technique; training data; Nonhomogeneous media; Robustness; Exchange rate prediction; Mean square error(MSE); Multi Layer Artificial Neural Network(MLANN); Wilcoxon Neural Network(WNN); Wilcoxon Norm(WN);
Conference_Titel :
Advances in Engineering, Science and Management (ICAESM), 2012 International Conference on
Conference_Location :
Nagapattinam, Tamil Nadu
Print_ISBN :
978-1-4673-0213-5
