Supervised classification of motor unit action potentials based on estimates of classification certainty

Abstract

The classifmtion of motor unit d o n potentials extracted from a myoelectric (ME) signal is a dif'fEcult task due to the lack d a priori information regarding the number of classes and the class statktics. Classification can be performed with great accuracy by humans due to inherent pattern analysis and decisionmaking abilities. In trying to match the performance of the human operator, a method was developed to combbe various sources of information into a certainty function which mimics the degree of certainty a human might have about a Classification, gken the same information. This method i9 described and prelimioary comparative results provided.