Fig. 1
The architecture of the VAE model. The encoder receives a DD map and outputs the parameters, μ and σ, which determine the distribution of the latent variable. The mean parameter, μ, and the variance parameter, σ, are employed to create a distribution for sampling the latent variables. The latent variable, z, is passed to the decoder via the sampling step, and the decoder reproduces the input image from the latent variables.