An Empirical Study of Factorization Methods to Quantify Motor Synergies

Abstract

This paper presents a study investigating robustness of three factorization methods in identifying muscle and kinematic synergies in healthy population. Results showed that Principal Component Analysis (PCA) and Non-Negative Matrix Factorization (NNMF) have comparable empirical performances, both outperforming Independent Component Analysis (ICA). However, PCA showed a faster training time, giving it an edge over NNMF.