Fig. 1
From: Gene2vec: distributed representation of genes based on co-expression
The Skip-Gram architecture was used for training for gene embedding. This is the modified architecture which is equivalent to the original word2vec, adopted from this blog [22]