Enhanced SDD Algorithm Optimization Technique for Finding Hyper parameter of SVM

Abstract

Background: It is crucial to pay attention to the classify data. The classification of data via Support Vector Machine (SVM) approach has severe restrictions. Corresponding to this, the intriguing improvements could not be accomplished without a suitable Support Vector Machine (SVM) classifier improvement and it is of high significance to build a machine learning model which can accurately classify the data. In this paper, an enhanced framework is proposed mainly used for classifying the data by introducing a hyperplane.

Objective: The most important aspect of this whole framework is to create an enhanced version of recently developed evolutionary algorithm known as Social Ski Driver (SSD) optimization. As far as we know, enhanced version of SDD optimization algorithm have not yet used in SVM hyperparameter optimization to classify data.

Methods: We, improvise Social Ski Driver (SSD) exploitation ability, with Levy flight. To verify this, the proposed method is then applied to balanced, imbalance and multiclass datasets with higher dimensionality from the UCI repository then empiercally compared with Grid Search, PSO and SSD-SVM.

Results: The result achieved shows that ESSD-SVM is capable of finding, optimal solution and better performance classification as compared with other approaches

Conclusion: The proposed ESSD-SVM model's effectiveness is demonstrated by its accuracy that indicates that it optimizes classification performance for hybrid models, which takes less time.