Comparative study of bagging, boosting and convolutional neural network for text classification Ramraj S1,*, Saranya S1, Yashwant K2 1Assistant Professor, Dept. of Software Engineering, SRM Institute Of Technology, Chennai. 2Student, Dept. of Software Engineering, SRM Institute Of Technology, Chennai. *Corresponding Author: Ramraj S Assistant Professor, Dept of Software Engineering, SRM Institute of Technology, Chennai Email: ramrajitsrm333@gmail.com
Online published on 16 October, 2018. Abstract Text classification is applied in various domains such as identifying malicious comments in a public forum, classifying documents according to its category, etc. Application of machine learning for the purpose of text classification is very successful in recent years, especially the algorithms based on bagging and boosting. Convolutional neural network (CNN), is one of the successful deep learning algorithm applied in many image classification problems. In this paper we evaluated the performance of tree based machine learning algorithms such as Random Forest, XGBoost along with SVM and CNN for text classification. Upon training we found that in terms of F1 score XGBoost was giving the best accuracy in long text classification while CNN were giving the best accuracy in short text classification. Top Keywords convolutional neural networks, decision tree ensemble, deep learning, ensemble, feature extraction, machine learning, support vector machines, XGBoost. Top |