Competitive coevolution for defense and security

Surrogate models of fitness have been presented as a way of reducing the number of fitness evaluations required by an evolutionary algorithm. This is of particular interest with expensive fitness functions where the cost of building the model is outweighed by the saving of using fewer function evaluations. In this paper we show how a Markov network model can be used as a surrogate fitness function in a genetic algorithm. We demonstrate this applied to a number of well-known benchmark functions and although the results are good in terms of function evaluations the model-building overhead requires a substantially more expensive fitness function to be worthwhile. We move on to describe a fitness function for feature selection in Case-Based Reasoning, which is considerably more expensive than the other benchmark functions we used. We show that for this problem using the surrogate offers a significant decrease in total run time compared to a GA using the true fitness function.

Hybridization of multi-objective evolutionary algorithms and artificial neural networks for optimizing the performance of electrical drives

Research on software protection based on virtual machine

A guide to NBS computer security literature

Computer Forensics is a research hot topic in the field of computer security with the recent increases in illegal accesses to computer system. According to the procedure of computer forensics, this paper presents the frame model of computer forensics, analyses the source of digital evidence. Because of digital feature, it is especially critical to how to secure the protection of digital evidence and make computer forensics have legal recognition of ability. From the computer evidence collection phase, transmission phase and the storage stage, this paper discusses the key technologies and approaches to ensure the security of digital evidence respectively. Through the guidance of the frame model and the security guarantee to each stage of computer forensics, the evidence would eventually be provided to the court.

Today’s age is Machine Learning (ML) and Data-Mining (DM) Techniques, as both techniques play a significant role in measuring vulnerability prediction accuracy. In the field of computer security, vulnerability is a fault that might be exploited as a risk artist that performs unlawful activities inside computer security. The attackers have several different fitting tools and they are taking advantage to operate software illegally and are using it for getting self-profit. Additionally, that helps to expose and identify the violence external. Weakness management remains a repeating exercise to identify, remediating, and justifying weaknesses. These exercises normally send software faults in computing security. The meaning of using weakness with the same risk might go to misperception. It is possible to have a major effect because of possible stability and the window of weakness presented a risk hole in the software and required to fruitfully finish and smoothly operate. A security room has to be set up (zero-day invaders). Software Security Faults stand serious among unavoidable complications in the realm of computer risk. In this study, we have provided a comprehensive review of three book chapters, more than a hundred research articles papers, and several associated papers of different work that have been studied within the capacity of SVA and discovery applying ML and data-mining techniques. The earlier work has been thoroughly read and an adequately comprehensive summary has been provided in table-1. ML techniques that can professionally handle these attacks and we expect the net result of this survey article to help indesigning the new detection model for identifying the above-mentioned attacks

Today, due to the evolution of technology and the use of the Internet on a large scale, securing everything is becoming an unavoidable necessity and a challenge for most companies. And since the traditional means of security have become insufficient due to the increase in the number and types of computer attacks that appear almost daily, researchers in the field of computer security are busy developing security tools based on artificial intelligence concepts to detect new attacks. In this work, we proposed a binary classification method for intrusion detection that has a high accuracy, precision and recall rates. This approach is based on multi-layer perceptron using both pearson correlation coefficient and simulated annealing for selecting attributes from the three datasets used for genarating and evaluating this model which are: NSL-KDD, UNSW-NB15 and CICIDS2017. We obtained 97,02% accuracy for NSL-KDD, 92,32% accuaracy for UNSW-NB15 and 97,70% for CICIDS2017.KeywordsIntrusion detectionMulti-layer perceptronPearson correlation coefficientSimulated annealing

Cyber-attacks are serious problems for the IT sector. According to a survey, 66 percent of businesses have experienced these attacks. It's difficult and tiresome to fend against these attacks. Cyber analysts are now studying hacker intentions and purposes for incidents in addition to analysing cyber occurrences as a result of the rise in cyberattacks. DBMS may also establish itself as a key technology in the field of computer security and addressing cyber threats among the several technologies being designed and developed to avoid cybercrime. Research on cybercrime and hacker's interpretation can be honed with the use of DBMS. Through its logical and physical models and query processing, it aids in supplying sufficient knowledge to understand upcoming cyber-attacks.

In the field of computer security, the intrusion detection system plays a major role. Whenever the intrusion takes place, intrusion detection technology is used to ensure the data integrity and system availability. Normally the intrusion is considered as an abnormal activity performed by the intruder. The breaking of information system or occurrence of any abnormal activity if performed by the intruder is known as intrusion. Based on the nature of access the intruders can be broadly classified into two groups, as internal intruders and external intruders. One who is having the access permission but trying to perform unauthorized or illegal activities is known as internal intruders. On the other hand one who do not have the access permission, but attacks the system with variety of techniques is called identified as external intruders.

Owing to the sharp rise in the severity of the threats imposed by software vulnerabilities, software vulnerability detection has become an important concern in the software industry, such as the embedded systems industry, and in the field of computer security. Software vulnerability detection can be carried out at the source code or binary level. However, the latter is more impactful and practical since when using commercial software, we usually only possess binary software. In this paper, we leverage deep learning and kernel methods to propose the Deep Cost-sensitive Kernel Machine, a method that inherits the advantages of deep learning methods in efficiently tackling structural data and kernel methods in learning the characteristic of vulnerable binary examples with high generalization capacity. We conduct experiments on two real-world binary datasets. The experimental results have shown a convincing outperformance of our proposed method over the baselines.

The Laboratory of Cryptography and System Security (CrySyS) is dedicated to conduct research in the field of computer security and user privacy. This paper shows a research roadmap of the CrySyS Lab from its inception in 2003 until today. We will present the major achievements in the past with a particular emphasis on the research and teaching curriculum in security and privacy. We will discuss network- and wireless system security, the core competences of CrySyS. Building on the research foundation and competences in these areas, we will lead the laboratory into new territories of security and privacy in wireless embedded computing systems and the future Internet.

The image files have data areas that are not very important for the visualization of the corresponding image and, if these areas are changed or altered, the image does not have a perceptible visual change, therefore they can be used to hide information, having almost certainty that the image has not been altered.The least significant bit (LSB) method consists in replacing the last bit of each byte and can do it in several bytes at the same time and the image will not have a noticeable change. Each substituted bit is part of a letter of the message that you want to hide in the given image. In general, images with PGM extension are very useful to apply the technique (LSB) since they are images whose content is given without following some algorithm of compression; This means that the content of the file is more readable and easy to work since it is in the ascci format (US standard code for information exchange).The content of an image with PGM extension, can be visualized using a simple text editor, so that the ASCII information can be seen as a representation of whole numbers, the value of these numbers will represent the intensity of color, 255 is white and 0 is black.If each row of the matrix represents an image data and every last digit, which is the least significant bit, is altered, then the eight bits corresponding to each altered byte would represent a letter belonging to a hidden message word. If this sequence is repeated several times you can hide more information such as phrases or sentences.Steganography is a useful tool in the field of computer security where it is required to verify that private information is correctly used by the user determined using the encryption in the personal key. There are already several software products that are responsible for carrying out this process, one of these cases is OpenStego, which encrypts the information using steganography and through passwords ensures that only the bearer of it will be able to visualize said information.We will show that the use of the Fortran, MatLab programming language presents great advantages for steganography since access to memory spaces through pointers allows freeing up memory at runtime, which provides a high degree of efficiency in the execution of a Program. Fortran, being a programming language oriented to mathematical calculation in general, has tools that facilitate its application and expressly the matrix calculation used in steganography.KeywordsSteganographyCybersecurityLSBPGMInformationEncryptionKey

The nature-inspired approaches and methods are widely used now for the solution of problems in cases when classical mathematical ones fail. Fully it concerns also the problems solved in the field of computer security. One of such types of problems is the optimization of access-control schemes in computer networks. A common example of such problem is the formation of access-control schemes in databases, which use a role-based access-control (RBAC) model. This problem was singled out in a separate direction in data mining and received the name "role mining problem" (RMP).Another example is to generate a scheme of the virtual local computer network (VLAN) where network hosts must be distributed among the minimum quantity of virtual subnets. Both problems are NP-complete. A set of different approaches was offered to solve these problems. However, none of them can be considered as universal one. At the same time, genetic algorithms, which are one of the most characteristic representatives of nature-inspired approaches and methods, can be considered as a basis for such universal solutions. The chapter outlines the mathematical foundations for generation of access-control schemes in computer networks, and examples of generating the RBAC and VLAN schemes are considered. The chapter demonstrates that these optimization problems belong to the class of Boolean matrix factorization (BMF) problems. Besides, it shows how to use genetic algorithms for their solving. Some enhancements in these algorithms which allow increasing the speed of their operation and accuracy of decisions are considered.

Measuring the usability and user performance level of any products or services especially in the field of online and computer-based applications is a vital issue, which has been in the centre of attention for the researchers. One of the most common ways of measuring the usability and performance of any online interactive application or service is by conducting a comprehensive user study or by asking users to give their feedback on that particular product or application. In order to achieve valuable user feedback, a good quality survey and questionnaire needs to be designed in order to measure the usability and performance of the model quantitatively and qualitatively. In this paper, some of the most important elements involved in designing of online survey forms especially in the field of computer security and precisely web security have been investigated in order to measure the usability and performance level of different online security applications, such as CAPTCHAs or Human Interactive Proofs (HIPs), which are used to increase the security of online-based service providers.

A byproduct benefit of using probabilistic model-building genetic algorithms is the creation of cheap and accurate surrogate models. Learning classifier systems -- and genetics-based machine learning in general -- can greatly benefit from such surrogates which may replace the costly matching procedure of a rule against large data sets. In this paper we investigate the accuracy of such surrogate fitness functions when coupled with the probabilistic models evolved by the $\chi$-ary extended compact classifier system ($\chi$eCCS). To achieve such a goal, we show the need that the probabilistic models should be able to represent all the accurate basis functions required for creating an accurate surrogate. We also introduce a procedure to transform populations of rules based into dependency structure matrices (DSMs) which allows building accurate models of overlapping building blocks -- a necessary condition to accurately estimate the fitness of the evolved rules.