Probabilistic Inference of Biological Networks via Data Integration

Table 3

Kernel weights learned for a comprehensive kernel, , that combines all base pairwise kernels. For each pairwise kernel, we show the final weight assigned to each of its base kernels. The tensor product kernel () and the metric learning kernel () contribute the most information to this comprehensive kernel. None of the motif base kernels () contribute, nor do any of the Cartesian product base kernels (). The kernel weights sum to unity.

Kernel	Kernel weights for combined model
		0	0.193	0	0.103	0.075	0.372
	0	0.002	0	0.010	0	0
	0	0.044	0	0.023	0	0.153
	0	0.006	0	0.019	0	0
	0	0	0	0	0	0