Using deep maxout neural networks to improve the accuracy of function prediction from protein interaction networks
Fig 2
scores obtained by network embedding representations and the corresponding STRING2GO-learnt functional representations based on (a) Combinedscore, (b) Textmining, (c) Experimental, (d) Database and (e) Coexpression networks by using SVM or Sigmoid function over the 10-fold cross validation during the GO term-specific classifiers training stage.