Fig. 7
From: Resolving the effect of wrist position on myoelectric pattern recognition control
Three classifiers were trained using three datasets: (1) unadjusted model-generated data which was a simulated dataset generated by randomly sampling from distributions described by the mean and variance generated by the neural network for only the neutral position; (2) adjusted model-generated data which was a simulated dataset generated by randomly sampling from distributions described by the means and variances generated by the neural network from all positions; and (3) the real data set. a Results from non-amputee subjects; for each muscle set, datasets that are significantly different from each other do not share the same symbol. b Results from two partial-hand amputee subjects. Error bars represent standard deviation