Ensemble Models for Cyber Attacks Detection on Multiple Datasets

Cyber attacks have become more complicated for banking, e-commerce, and other social media applications. Many traditional models can prevent threats or attacks from malicious online users. This paper developed a new deep learning (DL) based model that detects cyber attacks by providing cyber security for the applications and the enhanced detection and classification of attacks. DL algorithms mainly focus on processing large and complex datasets consisting of types of attacks from different online sources. Cyber security is the domain that process various tasks such as intrusion detection, classification of malicious activities, and detection of phishing attacks detection. This work developed the integrated approach Autoencoder and Recurrent Neural Networks (RNNs) that detect and classify the attacks. The researchers used the pre-trained DL-based Convolutional Neural Network (CNN) model to train on attack datasets to obtain accurate patterns from the datasets. They use preprocessing and feature extraction to improve the performance of the proposed A-RNN approach. The quantitative performances of algorithms show the detection and classification rate.