Fig. 4From: A multitask transfer learning framework for the prediction of virus-human protein–protein interactionsAblation study on benchmarked datasets. Compared with STT, MTT is statistically better in five datasets, while on the remaining four (Novel H1N1, Denovo_slim, Yersina, and Franci), the difference is not statistically significant. MTT is statistically better than Naive baseline on eight out of nine datasets, while on the remaining dataset(Novel Ebola), the difference is not statistically differentBack to article page