On the Use of Neuro-Fuzzy Techniques for Analyzing Experimental Surface Electromyographic Data

n this paper, the electrical signals coming from muscles in activity through experimental electromyogram interference patterns measured on human subjects are investigated. The experiments make use of surface ElectroMyoGraphic (sEMG). The use of Independent Component Analysis (ICA) is suggested as a method for processing raw sEMG data by reducing the ”cross-talk” effect. ICA also allows us to remove artefacts and to separate the different sources of muscle activity. The main ICs are used to reconstruct the original signal by using a neuro-fuzzy network. An auto-associative Neural Network that exploits wavelet coefficients as an input vector is also used as simple detector of non-stationarity based on a measure of reconstruction error.