Fig. 2From: Multi-objective de novo drug design with conditional graph generative modelA schematic representation of molecule generation process. Starting with the empty graph \(G_0\), initialization is performed to add the first atom. At each step, a graph transition (append, connect or terminate) is sampled and performed on the intermediate molecule structure. The probability for sampling each transition is given by \(p_{\varvec{\theta }}(t|G_i,\ldots, G_0)\), which is parametrized using deep neural network. Finally, termination operation is performed to end the generationBack to article page