Applications of Computational Intelligence in Computing Security: A Review

Introduction

The aim of artificial intelligence (AI) is to simulate human intelligence on machines so that they can act and think like humans. AI is regarded as a wide field of knowledge that involves reasoning, machine learning, planning, intelligent search and building perception. Reasoning aims to reach a predetermined objective of a problem using a set of facts supplied and a predefined base of information or what we call a knowledge base. The knowledge base consists of a compilation of IF-THEN rules or a conceptual graphic structure reflecting the knowledge of the professional in a specialized field. On the other hand, learning can be described as a process of encoding the situation-action pairs at memory so that the memory can remember the correct action. The learning process is carried out on machines either by training the machine with established situation-action pairs or by enabling the machine to adjust the parameters of the specified learning rule in a trial sense. Whereas preparation is about the sequence of steps to solve a problem. To be more precise, provided a knowledge base and a set of facts, preparation calls for the sequencing of the rule firing phase, so that there is at least one such target lading sequence.

Therefore, AI's main goals are to develop methods and systems for solving problems that are normally solved by human intellectual activity, such as image recognition, language and speech processing, preparation and forecasting, thereby enhancing computer information systems; and to develop models that simulate living organisms and, in general, the human brain, thereby improving our understanding. There is a large AI literature (Jain and Lazzerini, 199), (Jain, 1999) and (Mitchell, 1997) that covers various techniques of representation of information, reasoning (Shapiro, 2010) and (Rashmi & Neha, 2017)., machine learning (Shai & Shai, 2014) and (Yu-Wei, 2015), image and language understanding (Mishra, 2018) and (Rastgarpour & Shanbehzadeh, 2011), planning, smart search and realization of knowledge. A detailed discussion on these issues goes beyond this chapter's reach.

It is clearly reported by a number of researchers that AI was incompetent to meet the growing demand for search, optimization and machine learning in information systems with broad biological and commercial databases and factory automation for the steel, aerospace, energy and pharmaceutical industries claims Konar (2005). These pitfalls of traditional AI can be summarised as follows:

• •

  Traditional problem-solving approaches in AI are primarily concerned with the representation of problem states by symbols and the construction of a set of rules to define transitions in problem states.

• •

  In general AI is a tool with the capability to handle inductive and analogy-based learning, but is inefficient for supervised

• •

  Traditionally, AI is utilized functionally in search algorithm, but conventional AI is not very qualified to deal with real world optimization problems.

The shortcoming of this classical AI has opened up new opportunities for non-classical models in different intelligent based applications. Such computational analytical tools and techniques have led to a new field called computational intelligence.

Computational Intelligence (CI) is a set of computational models and techniques that incorporate components of learning, adjustment, or potentially heuristic advancement. It is utilized to help study issues that are hard to unravel utilizing regular computational calculations and algorithms. The three main pillars of CI are neural networks, evolutionary computing, and fuzzy systems. In recent years, the scope of computational intelligence technologies has been applied to emerging areas such as swarm intelligence, artificial immune systems (AIS), supporting vector machines, rough collections or sets, chaotic structures or systems, and others. Figure 1 illustrates the main pillars for CI.

Figure 1.

Computational intelligence pillars