Fig. 1
From: GraphTar: applying word2vec and graph neural networks to miRNA target prediction
High-level model overview: the input graph is created from the word2vec-encoded sequences. GNN layers process the graph and yield accurate node embeddings, resulting in a graph representation vector. The prediction head operator aggregates the embeddings, and finally, a set of fully connected layers classify the vector. Note that we provide dimension information: V denotes the number of nodes in the input graph, while D stands for the node embedding dimension, which is equal to the number of output channels of the last GNN layer