Genetic reasoning for finger sign identification based on forearm electromyogram

Abstract

This paper proposes a meta-heuristic data-clustering application to identify finger signs only by measuring surface electromyogram (EMG) of a forearm. It classifies EMG signal patterns peculiar to finger signs. Genetic programming learns intensity characteristics of EMG signals, and creates classification algorithm. Three typical finger signs are evaluated in terms of generated EMG. Experiments are conducted to reveal the successful identification of finger signs in real time.