Fig. 1From: Chemical toxicity prediction based on semi-supervised learning and graph convolutional neural networkThe SSL-GCN model for compound toxicity prediction. Molecular compounds are converted into graphs of nodes and connections. The GCN model architecture is composed of two stacked layers of graph convolutional layer, dropout, and batch normalization layer. All signals are summarized by the max pooling layer and fed into the multilayer perceptron network to generate the final output. The teacher and student GCN models are updated using the MT algorithmBack to article page