Fig. 2From: A Graph Feature Auto-Encoder for the prediction of unobserved node features on biological networksThe two different approaches of masking of expression values. In A, each dimension of \({\mathbf {X}}_{test}\) is predicted using a separate model trained on \({\mathbf {X}}_{train}\). In B, a single model is used to predict all the values of the feature matrixBack to article page