An ensemble approach applied to classify spam e-mails

Abstract

Spam e-mails, known as unsolicited e-mail messages, have become an increasing problem for information security. The intrusion of spam e-mails persecute the users and waste the network resources. Traditionally, machine learning and statistical filtering systems are used to filter out spam e-mails. However, there is no unique method can be successfully applied to classify spam e-mails. It is necessary to apply multiple approaches to detect spam and effectively filter out the increasing volumes of spam e-mails. In this paper, an ensemble approach, based on decision tree, support vector machine and back-propagation network, is applied to classify spam e-mails. The proposed approach is based on the characteristics of the spam e-mails. The spam e-mails are categorized into 14 features and then the ensemble approach is performed to classify them. From simulation results, the proposed ensemble approach outperforms other approaches for two test datasets.