The Improved LSTM and CNN Models for DDoS Attacks Prediction in Social Media

Abstract

Automatic identification of conversations related to DDoS events in social networking logs helps the organizations act proactively through early detection of negative and positive sentiments in cyberspace. In this article, the authors describe the novel application of a deep learning method to the automatic identification of negative and positive sentiments in large volumes of social networking texts. The authors present classifiers based on Convolutional Neural Network (CNN) and Long Short-Term Memory (LSTM) to address this problem domain. The improved CNN and LSTM architecture outperform the classification techniques that are common in this domain including classic CNN and classic LSTM in terms of classification performance, which is measured by recall, precision, f-measure, train loss, train accuracy, test loss, and test accuracy. In order to predict the occurrence probability of the DDoS events the next day, the negative and positive sentiments in social networking texts are used. To verify the efficacy of the proposed method experiments is conducted on Twitter data.