The effect of data reduction by independent component analysis and principal component analysis in hand motion identification

Both independent component analysis (ICA) and principal component analysis (PCA) were used in this study to evaluate their effects in data reduction in the hand motion identification using surface electromyogram (SEMG) and stationary wavelet transformation. The results indicate that both methods increase the number of training epochs of the artificial neural network. The unsupervised Fast ICA reduces the number of SEMG channels from 7 to 4. However the hand motion identification rate using the reduced channels is significantly lower (p < 0.05). On the other hand, the PCA reduces the size of neural network by more than 70%. Moreover, the results of discrimination rate and neural network training epochs show no significant difference as compared to the results before PCA reduction. The result of this study demonstrates that using wavelet and PCA are effective pre-processing for surface EMG analysis. It can efficiently reduce the size of neural network and increase the discrimination rate for different hand motions.