A Study of some ML and DL-based Strategies for Network Security
Abstract
The Internet and advanced communication networks, such as IoT and cellular net-
works, produce enormous and diverse traffic data flows. The behavior of network
traffic in these networks is highly intricate due to factors like device mobility and
network heterogeneity. As a result, conventional network security and management
methods struggle to handle the challenges of securing, monitoring, and analyzing
the network and data. These challenges include issues such as the efficacy of classi-
fication and detection strategies, precision, accuracy, and the ability to process big
data in real-time.
Recently, machine learning and deep learning have proven to be highly effective in
addressing network security concerns and have demonstrated superiority over tradi-
tional methods. Consequently, researchers in the field of networking are turning to
these machine learning and deep learning models for network security and manage-
ment.
Motivated by the success and effectiveness of these models, this thesis concentrates
on addressing two crucial and challenging issues in network security through dynamic
analysis, machine learning, and deep learning models, with a particular emphasis on
artificial neural networks. More precisely, it presents new learning-based techniques
for attack classification and malware detection.
First, a general introduction with some motivations for this thesis is presented.
Then, the state-of-the-art is investigated, and the most effective artificial intelligence-
based methods related to the above mentioned network security aspects are reported.
After that, the primary materials and preliminaries for constructing the proposed
models, such as neural network types, recurrence plots, and so on, are introduced
in detail. Finally, the proposed methods for attack classification and malware de-
tection are investigated in terms of mathematical background, model architecture,
experimental setting, and evaluation.
Moreover, the proposed methods are analyzed from a theoretical perspective and
through specific performance evaluation experiments on real network traffic datasets.
The obtained results, in the presence of several unbalanced datasets instances, prove
the effectiveness of the proposed approaches.
Keywords: Network Security, Attack Detection, Attack Classification, Mal-
ware Detection, Machine Learning, Deep Learning, Neural Networks, Recurrence
Plots, Markov Chain. [edited by Author]