Construction of a multi-agent attack detection system based on artificial intelligence models

2021 ◽  
Vol 26 (jai2021.26(1)) ◽  
pp. 22-30
Author(s):  
Belej O ◽  
◽  
Spas N ◽  
Artyshchuk I ◽  
Fedastsou M ◽  
...  
Keyword(s):  
Artificial Intelligence ◽  
Neural Networks ◽  
Information System ◽  
Detection System ◽  
Attack Detection ◽  
Sources Of Information ◽  
Joint Decision ◽  
Different Types ◽  
Artificial Intelligence Models ◽  
Multi Agent

Statistics of recent years on attacking actions on information systems show both the growth of known attackers and the growth of new models and directions of attacks. In this regard, the task of collecting information about events occurring in the information system and related to the main objects of the information system, and conducting their effective analysis is relevant. The main requirements for the tools of analysis are: speed and ability to adapt to new circumstances - adaptability. Means that meet these requirements are artificial intelligence systems. In particular, there are a number of research that use neural networks as a means of analysis. There are different types of neural networks, which differ depending on the tasks to be solved and are more suitable for different input data. The proposed multi-agent attack detection system collects and analyzes the collected information about the events of the information system using two types of neural networks. A multilayer perceptron is used to analyze various logs of information system objects. The Jordan network is used to analyze directly collected information about the events of information system objects. The use of a multi-agent attack detection system can increase the security of the information system. Features of modern attacks are considered. The urgency of the task of detecting attacks is substantiated. The peculiarities of the attack process were considered. The actions of attackers of different types at different stages of the attack are analyzed. It was shown which methods of detecting attacks should be used at different stages of the attack by an attacker. A model of a multi-agent attack detection system is proposed. An interpretation of the results of the analysis of information system events by the method of detecting attacks was proposed, as well as an algorithm for joint decision-making by agents based on several sources of information about their status. A model of an attack detection system that takes into account these features is proposed. This attack detection system collects information at several levels of the information system and uses it to analyze the artificial intelligence system

scholarly journals MEASUREMENT ERRORS AFFECTING THE CHARACTERISTICS OF MULTI-POSITION SYSTEMS, AND WAYS TO REDUCE THEM

InterConf ◽  
2021 ◽  
pp. 333-346
Author(s):  
Andriy Аrtikula ◽  
Dmytro Britov ◽  
Volodymyr Dzhus ◽  
Borys Haibadulov ◽  
Anastasiia Haibadulova ◽  
...  
Keyword(s):  
Information System ◽  
Information Security ◽  
Human Development ◽  
Measurement Errors ◽  
Information Needs ◽  
Single System ◽  
Sources Of Information ◽  
Different Types ◽  
System Errors ◽  
System Characteristics

Modern wide development of science and technology causes the growth of information needs in all branches of human development. At present, there are all opportunities to increase information security by combining sources of information into a single system. At the same time, when merging, specific difficulties and features emerge, which together make it difficult to implement the proposed solutions. The paper considers the peculiarity of combining different types of radar stations into a single information system. Errors of measurements of separate parameters and their influence on system characteristics are considered. Options for solving the problems that have arisen are proposed.

scholarly journals INTELLIGENT MULTI-AGENT INFORMATION SECURITY SYSTEM

2014 ◽  
pp. 35-39
Author(s):  
N. Kussul ◽  
A. Shelestov ◽  
A. Sidorenko ◽  
S. Skakun ◽  
V. Pasechnyk
Keyword(s):  
Neural Network ◽  
Neural Network Model ◽  
Intelligent Agents ◽  
Intrusion Detection System ◽  
User Behavior ◽  
Detection System ◽  
System Behavior ◽  
Agent Based ◽  
Different Types ◽  
Multi Agent

It is proposed an agent approach for creation of intelligent intrusion detection system. The system allows detecting known type of attacks and anomalies in user activity and computer system behavior. The system includes different types of intelligent agents. The most important one is user agent based on neural network model of user behavior. Proposed approach is verified by experiments in real intranet of Institute of Physics and Technologies of National Technical University of Ukraine "Kiev Polytechnic Institute.

scholarly journals RLXSS: Optimizing XSS Detection Model to Defend Against Adversarial Attacks Based on Reinforcement Learning

2019 ◽  
Vol 11 (8) ◽  
pp. 177
Author(s):  
Yong Fang ◽  
Cheng Huang ◽  
Yijia Xu ◽  
Yang Li
Keyword(s):  
Artificial Intelligence ◽  
Reinforcement Learning ◽  
Learning Algorithms ◽  
Attack Detection ◽  
Machine Learning Algorithms ◽  
Detection Model ◽  
Attack Model ◽  
Intelligence Models ◽  
Artificial Intelligence Models ◽  
Adversarial Model

With the development of artificial intelligence, machine learning algorithms and deep learning algorithms are widely applied to attack detection models. Adversarial attacks against artificial intelligence models become inevitable problems when there is a lack of research on the cross-site scripting (XSS) attack detection model for defense against attacks. It is extremely important to design a method that can effectively improve the detection model against attack. In this paper, we present a method based on reinforcement learning (called RLXSS), which aims to optimize the XSS detection model to defend against adversarial attacks. First, the adversarial samples of the detection model are mined by the adversarial attack model based on reinforcement learning. Secondly, the detection model and the adversarial model are alternately trained. After each round, the newly-excavated adversarial samples are marked as a malicious sample and are used to retrain the detection model. Experimental results show that the proposed RLXSS model can successfully mine adversarial samples that escape black-box and white-box detection and retain aggressive features. What is more, by alternately training the detection model and the confrontation attack model, the escape rate of the detection model is continuously reduced, which indicates that the model can improve the ability of the detection model to defend against attacks.

scholarly journals A2CM: a new multi-agent algorithm

ACTA IMEKO ◽  
2021 ◽  
Vol 10 (3) ◽  
pp. 28
Author(s):  
Gabor Paczolay ◽  
Istvan Harmati
Keyword(s):  
Artificial Intelligence ◽  
Neural Networks ◽  
Reinforcement Learning ◽  
Collision Avoidance ◽  
Multiple Agents ◽  
Pursuit Evasion ◽  
The World ◽  
Multi Agent ◽  
New Algorithms ◽  
Modified Algorithm

<p class="Abstract">Reinforcement learning is currently one of the most researched fields of artificial intelligence. New algorithms are being developed that use neural networks to compute the selected action, especially for deep reinforcement learning. One subcategory of reinforcement learning is multi-agent reinforcement learning, in which multiple agents are present in the world. As it involves the simulation of an environment, it can be applied to robotics as well. In our paper, we use our modified version of the advantage actor–critic (A2C) algorithm, which is suitable for multi-agent scenarios. We test this modified algorithm on our testbed, a cooperative–competitive pursuit–evasion environment, and later we address the problem of collision avoidance.</p>

scholarly journals A Residual Learning-Based Network Intrusion Detection System

2021 ◽  
Vol 2021 ◽  
pp. 1-9
Author(s):  
Jiarui Man ◽  
Guozi Sun
Keyword(s):  
Neural Networks ◽  
Intrusion Detection ◽  
Detection System ◽  
Attack Detection ◽  
Network Intrusion Detection ◽  
Detection Accuracy ◽  
Class Imbalance Problem ◽  
Data Set ◽  
Residual Learning ◽  
Network Intrusion

Neural networks have been proved to perform well in network intrusion detection. In order to acquire better features of network traffic, more learning layers are necessarily required. However, according to the results of the previous research, adding layers to the neural networks might fail to improve the classification results. In fact, after the number of layers has reached a certain threshold, performance of the model tends to degrade. In this paper, we propose a network intrusion detection model based on residual learning. After transforming the UNSW-NB15 data set into images, deeper convolutional neural networks with residual blocks are built to learn more critical features. Instead of the cross-entropy loss function, the modified focal loss is calculated to address the class imbalance problem in the training set and identify minor attacks in the testing set. Batch normalization and global average pooling are used to avoid overfitting and enhance the model. Experimental results show that the proposed model can improve attack detection accuracy compared with existing models.

scholarly journals Relating Blindsight and AI: A Review

2021 ◽  
pp. 1-15
Author(s):  
Joshua Bensemann ◽  
Qiming Bao ◽  
Gaël Gendron ◽  
Tim Hartill ◽  
Michael Witbrock
Keyword(s):  
Artificial Intelligence ◽  
Neural Networks ◽  
Computational Models ◽  
Visual Experience ◽  
Biological Research ◽  
Future Research ◽  
Network Architectures ◽  
Intelligence Models ◽  
Artificial Intelligence Models ◽  
Artificial Neural

Processes occurring in brains, a.k.a. biological neural networks, can and have been modeled within artificial neural network architectures. Due to this, we have conducted a review of research on the phenomenon of blindsight in an attempt to generate ideas for artificial intelligence models. Blindsight can be considered as a diminished form of visual experience. If we assume that artificial networks have no form of visual experience, then deficits caused by blindsight give us insights into the processes occurring within visual experience that we can incorporate into artificial neural networks. This paper has been structured into three parts. Section 2 is a review of blindsight research, looking specifically at the errors occurring during this condition compared to normal vision. Section 3 identifies overall patterns from Sec. 2 to generate insights for computational models of vision. Section 4 demonstrates the utility of examining biological research to inform artificial intelligence research by examining computational models of visual attention relevant to one of the insights generated in Sec. 3. The research covered in Sec. 4 shows that incorporating one of our insights into computational vision does benefit those models. Future research will be required to determine whether our other insights are as valuable.

scholarly journals Autonomous Interpretation of the Microstructure of Steels and Special Alloys

2019 ◽  
Vol 949 ◽  
pp. 24-31 ◽  
Author(s):  
Bartłomiej Mulewicz ◽  
Grzegorz Korpala ◽  
Jan Kusiak ◽  
Ulrich Prahl
Keyword(s):  
Artificial Intelligence ◽  
Neural Networks ◽  
Classification Systems ◽  
Training Data ◽  
Deep Convolutional Neural Networks ◽  
Data Set ◽  
Steel Microstructure ◽  
Different Types ◽  
Steel Grades

The main objective of presented research is an attempt of application of techniques taken from a dynamically developing field of image analysis based on Artificial Intelligence, particularly on Deep Learning, in classification of steel microstructures. Our research focused on developing and implementation of Deep Convolutional Neural Networks (DCNN) for classification of different types of steel microstructure photographs received from the light microscopy at the TU Bergakademie, Freiberg. First, brief presentation of the idea of the system based on DCNN is given. Next, the results of tests of developed classification system on 8 different types (classes) of microstructure of the following different steel grades: C15, C45, C60, C80, V33, X70 and carbide free steel. The DCNN based classification systems require numerous training data and the system accuracy strongly depend on the size of these data. Therefore, created data set of numerous micrograph images of different types of microstructure (33283 photographs) gave the opportunity to develop high precision classification systems and segmentation routines, reaching the accuracy of 99.8%. Presented results confirm, that DCNN can be a useful tool in microstructure classification.

scholarly journals Artificial Intelligence Models for Crime Prediction in Urban Spaces

2020 ◽  
Vol 8 (1) ◽  
pp. 1-13
Author(s):  
Ana Laura Lira Cortes ◽  
Carlos Fuentes Silva
Keyword(s):  
Artificial Intelligence ◽  
Neural Networks ◽  
Urban Areas ◽  
Urban Context ◽  
Criminal Records ◽  
Legal Information ◽  
Legal Advice ◽  
Intelligence Services ◽  
Artificial Intelligence Models ◽  
The Government

This work presents research based on evidence with neural networks for the development of predictive crime models, finding the data sets used are focused on historical crime data, crime classification, types of theft at different scales of space and time, counting crime and conflict points in urban areas. Among some results, 81% precision is observed in the prediction of the Neural Network algorithm and ranges in the prediction of crime occurrence at a space-time point between 75% and 90% using LSTM (Long-ShortSpace-Time). It is also observed in this review, that in the field of justice, systems based on intelligent technologies have been incorporated, to carry out activities such as legal advice, prediction and decisionmaking, national and international cooperation in the fight against crime, police and intelligence services, control systems with facial recognition, search and processing of legal information, predictive surveillance, the definition of criminal models under the criteria of criminal records, history of incidents in different regions of the city, location of the police force, established businesses, etc., that is, they make predictions in the urban context of public security and justice. Finally, the ethical considerations and principles related to predictive developments based on artificial intelligence are presented, which seek to guarantee aspects such as privacy, privacy and the impartiality of the algorithms, as well as avoid the processing of data under biases or distinctions. Therefore, it is concluded that the scenario for the development, research, and operation of predictive crime solutions with neural networks and artificial intelligence in urban contexts, is viable and necessary in Mexico, representing an innovative and effective alternative that contributes to the attention of insecurity, since according to the indices of intentional homicides, the crime rates of organized crime and violence with firearms, according to statistics from INEGI, the Global Peace Index and the Government of Mexico, remain in increase.

scholarly journals A comparison of activation functions in multilayer neural network for predicting the production and consumption of electricity power

2021 ◽  
Vol 11 (1) ◽  
pp. 163
Author(s):  
Abdulwahed Salam ◽  
Abdelaaziz El Hibaoui ◽  
Abdulgabbar Saif
Keyword(s):  
Artificial Intelligence ◽  
Neural Networks ◽  
Activation Function ◽  
Activation Functions ◽  
Paper Cover ◽  
Intelligence Models ◽  
Artificial Intelligence Models ◽  
Electricity Power ◽  
Multilayer Neural Networks ◽  
Power Utilities

Predicting electricity power is an important task, which helps power utilities in improving their systems’ performance in terms of effectiveness, productivity, management and control. Several researches had introduced this task using three main models: engineering, statistical and artificial intelligence. Based on the experiments, which used artificial intelligence models, multilayer neural networks model has proven its success in predicting many evaluation datasets. However, the performance of this model depends mainly on the type of activation function. Therefore, this paper introduces an experimental study for investigating the performance of the multilayer neural networks model with respect to different activation functions and different depths of hidden layers. The experiments in this paper cover the comparison among eleven activation functions using four benchmark electricity datasets. The activation functions under examination are sigmoid, hyperbolic tangent, SoftSign, SoftPlus, ReLU, Leak ReLU, Gaussian, ELU, SELU, Swish and Adjust-Swish. Experimental results show that ReLU and Leak ReLU activation functions outperform their counterparts in all datasets.

Sign in / Sign up

Export Citation Format

Share Document