Page 278 - Artificial Intelligence in the Age of Neural Networks and Brain Computing

3. Multiview Learning in Bioinformatics 271

(A) (B) (C) (D)

FIGURE 13.5
SNF approach. The SNF tool first constructs the patients’ similarity matrices for each data
type (A) then it computes the patient-by-patient similarity networks (B) the networks are
then integrated in an iterative approach (C) the patient clusters are evaluated with a
spectral clustering algorithm on the fused network (D).

FIGURE 13.6
The multiview genomic data integration methodology is composed of four steps. First, a
dimensionality reduction is performed by clustering the features. A prototype is extracted
for each cluster to represent it in the following steps. Second, the prototypes are ranked by
the patient class separability and the most significant ones are selected. Third, single-view
clustering methods are applied to each view to group patients and obtain membership
matrices. Fourth, a late integration approach is used to integrate clustering results.