A Pi Sigma neural network based electromyograph signal identification method

Objective: To introduce a new method to improve training speed and scale limitation in artificial limb. Method:A new efficient method for electromyograph(EMG) features identification based on the wavelet transformation was introduced, using a novel higher order network called the Pi Sigma network. Result:The method had good convergence properties and accuracy compared with conventional ones. Conclusion:EMG identification with Pi Sigma would be rather promising in the future development in EMG identification area. Author′s address\ Dept.of Biomedical Engineering , Shanghai Jiaotong University ,200030