Figure 2 | Scientific Reports

Figure 2

From: Learning fine-grained estimation of physiological states from coarse-grained labels by distribution restoration

Figure 2

Visualization of PD-sEMG dataset. Each sEMG sequence is annotated with a class label among the five classes defined by the internationally accepted MDS-UPDRS19 rating scale. The class label indicates the tremor severity where Class 5 represents severe and Class 1 represents normal. It is clear that the sEMG features change gradually from Class 1 to Class 5 corresponding to the fact that the physiological states are naturally continuous. The discrete classification labels are only coarse approximations of the real situation and cannot capture the smooth transitions between neighbouring classes.

Back to article page