Research Article
Multi-Instance Multilabel Learning with Weak-Label for Predicting Protein Function in Electricigens
Figure 1
The performance of MIMLwel on all two datasets with 80% weak-label ratios (W.L.R.) under different values of scaling factorμ when the fraction parameterα is fixed to 0.1 and different values of the fraction parameterα when the scaling factorμ is fixed to 1.0. The performance of MIMLwel reaches the perk in most cases by setting the scaling factorμ to 1.0 and the fraction parameter to 0.1.