The paper states that a learned two-dimensional AEVB encoder can project MNIST and Frey Face observations into a low-dimensional latent space; Figure 4 itself shows grids of decoded model outputs p_theta(x|z) at inverse-Gaussian-CDF-transformed latent coordinates, not plotted encoded datapoint projections. · CiteArk
The paper states that a learned two-dimensional AEVB encoder can project MNIST and Frey Face observations into a low-dimensional latent space; Figure 4 itself shows grids of decoded model outputs p_theta(x|z) at inverse-Gaussian-CDF-transformed latent coordinates, not plotted encoded datapoint projections.
来源：source-paper:PDF pp. 8-10; Section 5 'Visualisation of high-dimensional data', Appendix A, and Figure 4 caption
