Skip to main content
. 2020 May 3;20(9):2605. doi: 10.3390/s20092605

Figure 4.

Figure 4

The proposed DCGAN architecture. Based on 400 points sampled randomly from the real dataset, the generator generates 2000 points that simulate the behavior of real patient tremor. The discriminator tries to distinguish the real data from the generated data, and both networks are updated based on the combined losses from the classification.