When the traditional decision tree algorithm is applied to the field of network security analysis, due to the unreasonable property selection method, the overfitting problem may be caused, and the accuracy of the constructed decision tree is low. Therefore, this paper proposes a decision tree selection method based on improved manifold learning algorithm. The manifold learning algorithm maps the high-dimensional feature space to the low-dimensional space, so the algorithm can acquire the essential attributes of the data source. According to this, the problems of low accuracy and overfitting can be solved. Aiming at the traditional manifold learning algorithms are sensitive to noise and the algorithms converges slowly, this paper proposes a Global and Local Mapping manifold learning algorithm, and this method is used to construct a decision tree. The experimental results show that compared with the traditional ID3 decision tree construction algorithm, the improved method reduces 2.16% and 1.626% in false positive rate and false negative rate respectively.
Network security Decision tree Manifold learning algorithm
This is a preview of subscription content, log in to check access.
Kim, S.Y., Upneja, A.: Predicting restaurant financial distress using decision tree and AdaBoosted decision tree models. Econ. Model. 36(1), 354–362 (2014)CrossRefGoogle Scholar
Pai, P.F., Changliao, L.H., Lin, K.P.: Analyzing basketball games by a support vector machines with decision tree model. Neural Comput. Appl. 232, 1–9 (2016)Google Scholar
Azad, M., Moshkov, M.: Multi-stage optimization of decision and inhibitory trees for decision tables with many-valued decisions. Eur. J. Oper. Res. 263, 910–921 (2017)MathSciNetCrossRefGoogle Scholar
Ai, X., Wu, J., Cui, Z.: Broaden the minority class space for decision tree induction using antigen-derived detectors. Knowl.-Based Syst. 137, 196–205 (2017)CrossRefGoogle Scholar
Cicalese, F., Laber, E., Saettler, A.: Decision trees for function evaluation: simultaneous optimization of worst and expected cost. Algorithmica 79, 1–34 (2013)MathSciNetzbMATHGoogle Scholar