The diagram in Figure 3 shows the architecture of the 65-32-8-32-65 autoencoder used in the demo program. An input image x, with 65 values between 0 and 1 is fed to the autoencoder. A neural layer ...

Using the log of the variance helps prevent values from becoming excessively large. [Click on image for larger view.] Figure 2: Variational Autoencoder Architecture for the UCI Digits Dataset The key ...