| Input:executedDataM//the set of processed feature attributes |
| Output:generatedClassifier//the generated optimized classifier |
(1) | Construct fuzzyFactor = null;//calculate the distance between each sample and the class as a fuzzy factor to improve the classification accuracy |
(2) | Construct executedDefaultStep = q, executedSearchStep = null;//control search time and grid density |
(3) | Construct executedPenaltyParameter;//express the fault tolerance of the sample data when constructing the classification plane of SVM |
(4) | Construct executedOverfittingThreshold = f;//judge whether the penalty parameter is within the critical range |
(5) | representCandidateParameters();//use grid nodes to represent candidate parameters |
(6) | set the range of parameters to generate grids in different directions; |
(7) | for each sample i in executedDataM do |
(8) | Construct executedSearchStep = q.t;// the incremental step is t times the default step q |
(9) | constructTraverseSearch();//perform traversal search on all samples |
(10) | divide into i-dimensional parameter space among i parameters; |
(11) | if (executedPenaltyParameter(i) < executedOverfittingThreshold) then |
(12) | executedSearchStep = 2/q;// reduce the step size to increase the grid density for a more accurate search |
(13) | constructTraverseSearch();//perform traversal search on all samples |
(14) | else |
(15) | expand the search space and adjust the search direction; |
(16) | constructTraverseSearch();//perform traversal search on all samples |
(17) | end if |
(18) | panel = createClassificationHyperplane();// construct the corresponding classification hyperplane |
(19) | calculateDistance(M[i], panel);// calculate the distance between each sample node and the hyperplane as a fuzzy factor |
(20) | computeFeatureValidity(i);// calculate the feature i of each sample data, which has a feature validity, and determine the classification effect of each feature |
(21) | useRadialBasisKernel();// the kernel function has lower complexity and higher classification efficiency |
(22) | end for |
(23) | generateClassification();// generate the optimized classifier |
(24) | return generatedClassifier; |