Department of Computer Engineering, Gujrat Technological College,
Ahmedabad, India
Email address:
Shezi1131@gmail.com
To cite this article:
Diptiben Ghelni. Deep Learning and Artificial Intelligence Framework to
Improve the Cyber Security. American Journal of Artificial
Intelligence . Vol. x, No. x, 2022, pp. x-x.
Abstract: Deep learning derived from an artificial neural
network (ANN), is one of the essential technologies for today’s
intelligent cyber security systems or policies. The benefits and
drawbacks of using artificial intelligence (AI) in cyber risk analytics
to improve organizational resilience and better comprehend cyber risk.
Multilayer perceptron, convolutional neural network, recurrent neural
network or long short-term memory, self-organizing map, auto-encoder,
restricted Boltzmann machine, deep belief networks, generative
adversarial network, deep transfer learning, and deep reinforcement
learning, as well as their ensembles and hybrid approaches, can be used
to tackle the diverse cyber security issues intelligently. The
backpropagation algorithm’s ultimate goal is to correctly maximize the
network weights to translate the inputs to the intended outputs. During
the training phase, several optimization approaches such as Stochastic
Gradient Descent (SGD), Limited Memory BFGS (L-BFGS), and Adaptive
Moment Estimation (Adam) are applied. These neural networks may be
utilized to handle a variety of cybersecurity problems. MLP-based
networks are used to construct an intrusion detection model, malware
analysis, security threat analysis, identify malicious botnet traffic,
and build trustworthy IoT systems. MLP is sensitive to feature scaling
and requires tuning a variety of hyperparameters such as the number of
hidden layers, neurons, and iterations, which might make solving a
complicated security model computationally costly.
Keywords: Cyber Security, Artificial Intelligence, Deep
Learning, Internet of Things
1. Introduction
Industry 4.0, an IoT phrase coined in 1999, is built on the Internet of
Things (IoT) technology, providing the first glimpse of what an
IoT-based ecosystem would look like in the future. CPS refers to the
interdisciplinary and complex characteristics of intelligent systems
constructed and relies on the interplay of physical and computational
components. CPS theory evolved from control theory and control systems
engineering. It focuses on the connectivity of physical features and the
utilization of sophisticated software entities to create new network and
system capabilities. CPSs connect biological and engineering systems,
bridging the cyber and physical worlds [1].
On the other hand, IoT theory is based on computer science and Internet
technologies, and it focuses primarily on the interconnection,
interoperability, and integration of physical components on the
Internet. This integration effort is expected to lead to advances such
as IoT automation of CPSs as the IoT industry matures over the next
decade. CPS systems and automated CPSs guide trained employees in
production situations in real-time. In this context, we look at how such
systems enable artificial intelligence (AI) breakthroughs in real-time
processing, sensing, and actuation across these new systems and give
cyber structure system analysis capabilities. As a result, we’ll
concentrate on artificial intelligence, which is a notion that
encompasses both the cyber-physical and social components of the hazards
associated with new technology deployment [2].
There are two research aims in this study. To begin, we provide an
up-to-date summary of current and emerging cyber risk analytics
breakthroughs. This incorporates current standards into a new risk
analytics feedback loop by combining existing literature to generate
shared core terminology and techniques. Second, by providing a novel
understanding of cyber network risk and the role of AI in future CPS, we
capture best practices and spark debate among practitioners and
academia. Throughout the article, this architecture is explored and may
be used as a best practice for designing and prototyping AI-enabled
dynamic cyber risk analyses [3].
2. Artificial intelligence, CPS, and predictive cyber risk analytics
literature review
The IoT has been defined as a revolutionary technological augmentation
that transforms the traditional living into a high-tech lifestyle in
terms of data streams. CPSs and IoT generate massive amounts of data,
necessitating powerful analytical tools for analysis. We almost likely
need AI-assisted analytical tools to clean up the data’s noise and
inconsistencies. On the other hand, CPS architectures cover a wide range
of topics. These many notions must be integrated into a system [4].
Furthermore, CPS mandates anti-counterfeiting and supply chain risk
management to combat malicious supply chain components that have been
altered from their original design to create disruption or perform
illegal functions. Hyper-connectivity in the digital supply chain must
be promoted in addition to design and process standardization. It is
proposed that restricting source code access to critical and experienced
employees can offer software assurance and application security and may
be required to prevent the introduction of purposeful faults and
vulnerabilities in CPSs. Forensics, prognostics, and recovery plans
should be included in security measures for cyber-attack analysis and
coordination with other CPSs and entities that detect external
cyber-attack vectors [5]. An internal track and trace network
procedure can help by recognizing or avoiding gaps in logistical
security measures. To prevent the exploitation of CPS vulnerabilities
discovered by reverse engineering assaults, a method for anti-malicious
and anti-tamper system engineering is required. Taxonomic analysis was
performed using the Smart literature review approach based on latent
Dirichlet allocation. The resulting areas of concentration are organized
into a taxonomy with acronyms to aid in the integration of artificial
intelligence with the current CPS. Deep learning (DL) is a subset of
machine learning (ML) and artificial intelligence (AI), and it is one of
the primary technologies of the Fourth Industrial Revolution. It is
derived from an artificial neural network (ANN) (Industry 4.0) [6].
”Cyber security” and ”Deep learning” are becoming increasingly popular
worldwide, as demonstrated in Figure 1.