Comparison of Deep and Traditional Learning Methods for Email Spam Filtering

Electronic mail, or email, is a method for com-municating using the internet which inexpensive, effective, and fast. Spam type of email where unwanted messages, usually commercial are distributed in large quantities by spammer. The objective such behavior to harm users; these messages need be detected prevented from being sent users first place. In order filter emails, developers have used machine learning methods. This paper discusses different methods deep as Convolutional Neural Network (CNN) Long Short-Term Memory (LSTM) models with(out) GloVe model classify spam non-spam messages. These only based on data, extraction set features automatic. addition, our work provides comparison between traditional algorithms datasets find out best way intrusion detection. results indicate that offers improved performance precision, recall, accuracy. As far we aware, show great promise able spam, therefore performed various with Using benchmark dataset consisting 5,243 16,872 not-spam SMS highest achieved accuracy score 96.52% CNN model.