Table 2 Summary and description of all the tested methods with all the tuned hyperparameters
From: Supervised multiple kernel learning approaches for multi-omics data integration
Methods | Integration | Optimized Parameters | Description |
|---|---|---|---|
block PLSDA | Mixed | ncomp | DIABLO |
block sPLSDA | Mixed | ncomp, keepX | DIABLO |
SVM concat | Early | C, \(\sigma\) | Direct concatenation |
SVM naive | Mixed | C, \(\sigma\) | Sum of the kernel |
SimpleMKL-SVM | Mixed | C, \(\sigma\) | Weighted sum of kernels |
SEMKL-SVM | Mixed | C, \(\sigma\) | Weighted sum of kernels |
STATIS-UMKL + SVM | Mixed | C, \(\sigma\) | Weighted sum of kernels |
Deep MKL | Mixed | \(\sigma\), epochs, principal components, dropout value | Deep Learning kernel fusion |
Cross-Modal Deep MKL | Mixed | \(\sigma\), epochs, principal components, dropout value | Deep Learning kernel fusion |
NN_VCDN | Late | NA | Feedforward neural network |
Dynamics | Late | NA | Dynamical Multimodal Fusion |
MOGONET | Late | Optimized k | Graph convolutional network |