Fig. 2
From: Multiple-kernel learning for genomic data mining and prediction
Results from SEMKL, SimpleMKL, and DAMKL on 9 benchmark datasets, where two radial kernels K1 and K2 with σ1=2 and σ2=0.05 were used. a Displays the learned kernel weight of K1 as the mean of each group changes. b Displays the predictive accuracy of each algorithm as the distance between each group changes. DAL Hinge and DAL Logistic refer to conducting DALMKL under different loss functions