For each pair of classes, a SVM decision function is trained for separating the two specific classes.
PPPE algorithm is a popular multiclass classification method that combines all comparisons for each pair of classes.
We believe the cause is that multi-class SVM [6] constructed multiple binary classifiers (e.g., K(K−1)/2 classifiers for K classes), one for each pair of classes, and finally identified a class based on majority voting.
Figure 6 shows a set of interaction networks for each pair of classes and within each class, corresponding to the entire monitoring period.
SVMs are binary classifiers and multi-class classification is realized by developing an SVM for each pair of classes.
Table 2 summarizes the number of NMJs and the corresponding data points used for comparing each pair of classes.
The differences between each pair of the classes were significant: p < 0.001 between class II and III, p = 0.002 between class III and IV, and p < 0.001 between class II and IV, respectively.
However, the post hoc comparison (by weighted Bonferroni test) showed highly significant differences between each pair of fibrosis classes for FibroMeter2G, whereas this was not observed between several pairs of contiguous classes of Fibrotest.
To test this hypothesis, they counted the number of paralogous gene pairs connecting each pair of gene classes (transcriptional classes, biochemical classes, etc).
Subsequently the relative risk was calculated between each pair of treatment classes.
Here is our rule of naming, for gene y, its median across samples 1 through xk in class k, then it will be named g y_median_C k, so g2_median_C2 means median value of gene 2 across samples in class 2. For each gene, after sorting the median values from smallest to largest, the fold-change of this gene between each pair of neighbor classes forms a stair, which can be diagrammed as below.
