Fig. 2From: A multitask transfer learning framework for the prediction of virus-human protein–protein interactionsComparison between MTT and state-of-the-art methods on small testing datasets. MTT is statistically better than Denovo in all benchmarked datasets. Compared with Generalized, MTT has higher performance in six out of seven datasets. MTT outperforms Doc2vec in four out of seven datasetsBack to article page