Fig. 18From: Classification complexity in myoelectric pattern recognitionData analysis tool using CCEAs. Data analysis module integrated into the open-source project BioPatRec [15]. A demonstration case shows data from a recording session of 11 classes (movements). Section 1 shows the feature space per pair of features (2D) for each class and closest neighbors. Section 2 shows the number of times that a movement is the most conflicting neighbor for any of the other movements. The displayed information is derived from the SI (either Bhattacharyya distance or the modified Mahalanobis as distance definitions), or by NNS. This tool provides useful information to identify classes that should be re-recorded or eliminated in order to improve the separability of the othersBack to article page