Skip to main content

Table 3 Comparison of different data integration algorithms

From: A comparison of graph- and kernel-based –omics data integration algorithms for classifying complex traits

Integration Algorithms

Computation Time

Stability

Characteristics

Graph-based semi-supervised learning

Low

Medium

Tuning parameter; performance can be poor sometimes

Graph sharpening integration

Low

Low

Tuning parameter; average weights frequently occur

Composite association network

Low

High

Average weights occur when all weights are negative

Bayesian network

Low

Low

Bins selection and training sample size affect performance

Semi-definite programming SVM

Medium

Low

Two tuning parameters; C is very sensitive to outliers

Relevance vector machine

High

High

Long training time; Probabilistic result

Ada-boost relevance vector machine

High

Medium

Resampling size and iteration can be hard to determine