Fig. 4From: Exploration and augmentation of pharmacological space via adversarial auto-encoder model for facilitating kinase-centric drug developmentVisualization by t-SNE of the distributions of generated sample, training set and test set. a Changes of data distribution during training of PCM-AAE. b Distribution of the training data for NB model. c Distribution of training set augmented by PCM-AAE. d Distribution of generated data from different generators trained with AAEBack to article page