Classification of Microarray Gene Expression Data Using an Infiltration Tactics Optimization (ITO) Algorithm

. 2020 Jul 18;11(7):819. doi: 10.3390/genes11070819

Algorithm 1: ITO Algorithm

input:

T: t × f matrix - training dataset with t samples and f features;

V: v × f matrix - validation dataset - with v samples and f features;

preps={Imputer, Robust, Quantile, Standard, …} - set of preprocessing methods;

searchRadius={10, 50, 100, 150, 200, 250, …} - set of FSS sizes;

searchStrategy={JMI, JMIM, mRMR …} - set of FSS methods;

successEvaluation={10 Fold CV, LOOCV, …} - set of validation methods;

L I G_{O p t i o n s}

={DT, AdaBoost, Extra Tree, …} - set of parameter-free classifiers;

F T_{O p t i o n s}

={DNN, SVM, Random Forest, …} - set of parameterized classifiers

output:

E n s e m b l e_{F i n a l}

BEGIN

G ← GenerateOptionsGrid(searchRadius, searchStrategy, successEvaluation, preps);

Choose

t_{L I G}

⊂ G using Randomized Grid Search;

L I G_{E n s e m b l e}

←

C o m p u t e L I G_{E n s e m b l e}

(T, V,

L I G_{O p t i o n s}

t_{L I G}

) //Algorithm 2;

Choose

t_{F T}

⊂ G using Randomized Grid Search;

E n s e m b l e_{F i n a l}

←

C o m p u t e E n s e m b l e_{F i n a l}

(T, V,

F T_{O p t i o n s}

t_{F T}

) //Algorithm 3;

return

E n s e m b l e_{F i n a l}

;

END