Indian Journal of Science and Technology
Year: 2023, Volume: 16, Issue: 33, Pages: 2601-2608
R Aravind Sekhar1*, K G Sreeni1
1College of Engineering, Trivandrum, India
Received Date:12 January 2023, Accepted Date:08 July 2023, Published Date:01 September 2023
Objectives: To develop an alternative approach to improve the performance of ensemble classifiers. Methods: The Principle of Maximum Entropy and a precision based diversity measure are used to develop the model. Here joint entropy values are used which strongly reflect the combined uncertainties in the actual prediction and the prediction obtained during the training phase. Towards that purpose probabilistic confusion matrix has been used. The performance of the proposed method is tested against twenty different datasets. Each dataset is different from one another in terms of the number of classes and features. The results have been compared with several existing state of art ensemble methods uch as Bagging, AdaBoost, Naive Bayes (NB), Weighted Majority Vote (WMV) and TransEnsemble Classifier (TrEnL). Findings:The results show that the proposed method can select the best among the classifiers and perform better when compared to conventional classifiers. Novelty: The maximum entropy principle is applied at the final prediction probability matrix so that the method can be used as a generalized technique to improve the performance of ensemble classifiers.
Keywords: Ensemble; Maximum Entropy; Confusion Matrix; Weighted Entropy; Probabilistic Confusion Matrix
© 2023 Sekhar & Sreeni. This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited. Published By Indian Society for Education and Environment (iSee)
Subscribe now for latest articles and news.