Fig. 8
From: Protein function annotation based on heterogeneous biological networks
ROC curves of eight methods using ten-fold cross validation. This Figure shows the ROC curves of PHN and other seven methods using ten-fold validation. Proteins are randomly divided into ten equal sets, one set for testing, and the rest for training. And then, the cross validation is repeated process for ten rounds, using various folds as the training and validation data in each round. The results of ten folds are averaged to generate the final performance. a Biological process, b molecular function, c cellular component