A Hybridized Deep Stacked Autoencoder Model for Botnet Detection
As the Internet aims to integrate and connect anything at anytime, anyplace with anything and by anyone,
cyberattacks develop in volume and complexity with Botnets which are used in a wide range of malicious
activities such as e-mail spamming; Phishing, social engineering, and even distributed denial of service
(DDoS) attack. This incessant increase of attacks necessitates the interests in detecting and preventing
botnet attacks in network and internet-based systems. This study develops a hybridized Deep Stacked
Autoencoder optimized with Genetic Algorithm (DSAE-GA) for the classification and identification of
intrusions from the internet environment. The hybridized DSAE-GA model primarily used Principal
Component Analysis (PCA) technique to select a subset of features. The DSAE was trained to learn the
normal network traffic profile using the Context Computer Network Traffic (CCNT) dataset while
adapting to reconstruct these points with minimal reconstruction error (RE). The design of the GA
majorly focuses on the parameter optimization of the DSAE thereby enhancing the classifier results.