A Boosted Text Classifier Based On Genetic Algorithm

In the field of text classifying, the GA based Support vector machine multiclassify is a widely used technology, it takes the advantage of self-searching of GA and excellent performance in high dimension classify of SVM, making it a appropriate approach in this area. Aiming at the defect in global optimization of the SVM Decision Tree based on Genetic Algorithm(GA-SVM), this paper redefined the fitness function,which is the key component in Genetic Algorithm, proposing an optimized cumulative fitness and a new SVM Decision Tree based on Cumulative Fitness Genetic Algorithm(CFGA-SVM). This algorithm took advantage of global optimization by the cumulative fitness in the gene selection phase of GA, thus bringing a more appropriate population into the following inheritance operation. Experimental results on the real Characters dataset verify that BCFGA-SVM is better than the traditional GA-SVM in the aspect of classification accuracy and global optimization measurement. It also has better performance in time complexity than CFGA-SVM due to the modified kernel parameter. BCFGA-SVM has wide application prospects especially with huge training sample.