Recognition of Multiple Imbalanced Cancer Types Based on DNA Microarray Data Using Ensemble Classifiers

Pseudocode 1

Pseudocode description of the FSS generation algorithm.

Input: training set 𝑇; Feature set 𝐹; Size of feature space 𝐾; Size of

feature subspace 𝐷; Number of feature subspace 𝐿

Output:  𝐿 feature subspace training subsets

Process:

(1) Gather features of 𝐹 into 𝐾 clusters by hierarchical clustering

  based on PCC: Cluster_𝑖 (1≤𝑖≤𝐾);

(2) For 𝑖=1 : 𝐾

(3) {

(4)  Select representative gene 𝑓_𝑖 in Cluster_𝑖 by SNR;

(5) }

(6) Construct feature space FS including all representive genes

   extracted above;

(7) For 𝑖=1 to 𝐿

(8) {

(9)   FSS_𝑖=𝑃FS∈𝑅^𝐾∈𝑅^𝐷;

(10)   𝑇_𝑖=(FSS_𝑖,𝑇); /*FSS: feature subspace

(11) }

(12) Output 𝐿 feature subspace training subsets