Academia.eduAcademia.edu

Figure 2 – uploaded by Gormanukonda Ravi Kumar

See full PDFdownloadDownload figure
Figure 3: compares the genetic-SV M approach with plain SVM classifier. This paper proposed a Genetic algorithm based optimization algorithm, which can optimize the parameter values for SVM, and obtain the optimal subset of features. A comparison of the proposed algorithm results with Existing SVM approach demonstrates that the proposed method improves the classification accuracy rates. The GA-SVM method was applied to remove insignificant features and effectively find best parameter values. The goal of this paper is to design Support V ector Machine and Binary Coded Genetic Algorithm were analyzed to find the classification accuracy and runtime for various kernel functions such as Polynomial and Radical Basic function are used. Feature Selection algorithm is used to improve the classification accuracy of classifier with respect to medical datasets. The results show that the classification accuracy of GA-SVM is the highest of SVM algorithm.

Figure 3 compares the genetic-SV M approach with plain SVM classifier. This paper proposed a Genetic algorithm based optimization algorithm, which can optimize the parameter values for SVM, and obtain the optimal subset of features. A comparison of the proposed algorithm results with Existing SVM approach demonstrates that the proposed method improves the classification accuracy rates. The GA-SVM method was applied to remove insignificant features and effectively find best parameter values. The goal of this paper is to design Support V ector Machine and Binary Coded Genetic Algorithm were analyzed to find the classification accuracy and runtime for various kernel functions such as Polynomial and Radical Basic function are used. Feature Selection algorithm is used to improve the classification accuracy of classifier with respect to medical datasets. The results show that the classification accuracy of GA-SVM is the highest of SVM algorithm.

Related Figures (3)

the case of examples not linearly separable, SVM uses a kernel functions to map the examples from input space into hi high dimensional feature space. Using a kernel function can solve the non-linear problem.
the case of examples not linearly separable, SVM uses a kernel functions to map the examples from input space into hi high dimensional feature space. Using a kernel function can solve the non-linear problem.
This section describes the experimental results obtained by applying the proposed algorithms to a variety of data sets. For experimentation, five benchmark datasets (Breast cancer, Pima Diabeties, Mammographic Mass, Dermatology and Thoracic Surgery data sets) are taken from the UCI machine learning repository[9] as shown in Table 1. In order to validate the prediction results of the comparison of the two classification(SVM, SVM + Genetic) techniques and the 10- fold crossover validation is used. The k-fold crossover validation is usually used to reduce the error resulted from random sampling in the comparison of the accuracies of a number of prediction models. The present study divided the data into 10 folds where 1 fold was for testing and 9 folds were for training for the 10-fold crossover validation. We have used the W eka toolkit to experiment with these five data mining algorithms [13]. [he Weka is an ensemble of tools for data classification, regression, clustering, association rules, and visualization. WEKA version 3.6.9 was utilized as a data mining tool to evaluate the performance and effectiveness of the SVM and Proposed SVM-Genetic technique. This is because the WEKA program offers a well defined framework for experimenters and developers to build and evaluate their models. The performance of a chosen classifier is validated based on error rate and computation time. The classification accuracy is predicted in terms of Sensitivity and Specificity. The computation time is noted for two classifier is taken in to
This section describes the experimental results obtained by applying the proposed algorithms to a variety of data sets. For experimentation, five benchmark datasets (Breast cancer, Pima Diabeties, Mammographic Mass, Dermatology and Thoracic Surgery data sets) are taken from the UCI machine learning repository[9] as shown in Table 1. In order to validate the prediction results of the comparison of the two classification(SVM, SVM + Genetic) techniques and the 10- fold crossover validation is used. The k-fold crossover validation is usually used to reduce the error resulted from random sampling in the comparison of the accuracies of a number of prediction models. The present study divided the data into 10 folds where 1 fold was for testing and 9 folds were for training for the 10-fold crossover validation. We have used the W eka toolkit to experiment with these five data mining algorithms [13]. [he Weka is an ensemble of tools for data classification, regression, clustering, association rules, and visualization. WEKA version 3.6.9 was utilized as a data mining tool to evaluate the performance and effectiveness of the SVM and Proposed SVM-Genetic technique. This is because the WEKA program offers a well defined framework for experimenters and developers to build and evaluate their models. The performance of a chosen classifier is validated based on error rate and computation time. The classification accuracy is predicted in terms of Sensitivity and Specificity. The computation time is noted for two classifier is taken in to
Related topics:
Artificial IntelligenceMachine Learning
580 California St., Suite 400
San Francisco, CA, 94104
© 2025 Academia. All rights reserved