Chair of Intelligent Systems and Networks

Scientific research conducted at the Department of Intelligent Systems and Networks is focused on the wide use of modern computer science and its related sciences in industry, management, applied research, and administration. A significant part of the obtained effects has already been verified in production conditions (industrial plants, local government units), where the high effectiveness of the proposed solutions has been confirmed. Besides that, the employees of the Department seek new methods in the field of artificial neural networks and machine learning that may be used in those areas. Research is conducted in several areas:

  • Cybersecurity
  • Analysis and synthesis of environmental and transport systems
  • Machine learning in intelligent systems
  • New architectures for neural networks

Contact: jkorniak@wsiz.edu.pl

Scientific research projects financed from external sources:


A new approach to effective training of complex intelligent systems
Project manager: Prof. Bogdan Wilamowski, Ph.D. Eng.
Successful completion of the project can solve many scientific and practical problems by replacing the traditional design approach with a new learning approach. This alternative method can have a broader meaning by finding solutions to many problems that until now were impossible to solve with traditional methods.
Period of implementation: 20.01.2016–25.06.2019
Financing: National Science Center OPUS program
Contact: jkolbusz@wsiz.edu.pl

Intelligent non-linear systems with shallow and deep architectures
Manager: Prof. Bogdan Wilamowski, Ph.D., Eng.
Recent research shows that the most popular architectures such as SLP (Single-Layer Perception) (MLP with one hidden layer) have very limited capabilities. For example, with a network of 10 SLP neurons, you can solve a Parity-9 problem while a FCC (Fully Connected Cascade) network with the same number of neurons allows to solve Parity-1023 problems. Unfortunately, popular learning algorithms (including the LM algorithm) are not capable of learning this type of compact and powerful architectures. The problems associated with the use of traditional neural networks pushed scientists to look for other directions, such as fuzzy systems, SVM (Support Vector Machine) or ELM (Extreme Learning Machines). It turns out that these complex problems can be solved using new compact architectures. Therefore, research in the project focused on networks with new compact architectures and new learning algorithms.
Period of implementation: 16.07.2014–15.01.2017
Financing: National Science Center OPUS program
Contact: jkolbusz@wsiz.edu.pl

Scientific research projects financed from internal sources:


Cybersecurity
Research team leader: Mirosław Hajder, Ph.D., Eng. mhajder@wsiz.edu.pl
Cooperation: employees of the AGH University of Science and Technology in Kraków, Rzeszów University of Technology, and the Vistula Academy of Finance and Business in Warsaw.
Cybersecurity of both classical information systems and critical infrastructure systems. The research includes:

  • Methods of automatic quality verification of security policies used in enterprises and administration. They result in tools supporting the creation of a technically and organizationally safe environment for processing classified information;
  • Automation of the threat detection process in information systems and methods of counteracting them, especially in cyber-physical systems. Unlike other works in that area, the proposed solutions are based on IoT components, and the computing power used is based on edge computing technology;
  • Security of industrial IT systems, in particular based on the Internet of Things and distributed computing in the context of solutions characteristic of the Industry 4.0 concept. In the solutions developed and created at present, special emphasis was placed on ensuring minimum sensitivity of products to damage to components and to hacker attacks;
  • Unattended staff training in the field of information security. As part of the research, modifications of modern remote teaching methods are developed to ensure maximum effectiveness of education and accuracy of knowledge assessment without direct, personal participation of a teacher.

Analysis and synthesis of environmental and transport systems
Research team leader: Mirosław Hajder, Ph.D., Eng. mhajder@wsiz.edu.pl
Cooperation: employees of the AGH University of Science and Technology in Kraków, Rzeszów University of Technology, and the Vistula Academy of Finance and Business in Warsaw.
Applied research is carried out in the following areas:

  • Analysis and synthesis of regional environmental monitoring systems. The work is focused on cyber-physical threat monitoring systems that use measurement data from the inhabitants of the monitored area. This unique solution allows to improve the accuracy of the forecast and increase the lead time in which it will appear. The forecasting uses methods of artificial intelligence, graph theory, and spectral analysis, among others.
  • Research on transport systems including i.a. winter support for maintenance-free road accessibility, focused on ecology, based on the methods and means of the Industrial Internet of Things. Within a given group of topics, work is also carried out on the optimization of passenger flows in autonomous communication systems on demand;
  • Construction is based on vital, fault-tolerant edge computing technologies, fog computing, and their implementation in industry and administration. The results of the research are verified on an ongoing basis in critical infrastructure systems, in which high availability, automatic restoration, and maintenance-free character are the prime features.

Recognition of multispectral input images using artificial dynamic neural networks
Project leader: Prof. Roman Peleshchak, Ph.D.
Coordinator: Jan Kopka
Team: Janusz Korniak, Ph.D., Eng.; Paweł Różycki, Ph.D., Eng.; Mariusz Wrzesień, Ph.D., Eng.;
Ph.D. student Ivan Peleshchak; Janusz Kolbusz, Ph.D., Eng. jkorniak@wsiz.edu.pl
The project aims to develop the architecture of self-developing artificial neural networks with dynamic neurons and their learning algorithm based on evolving modeling methods to recognize multispectral input images at a high speed of their reception. In particular, the work focuses on:

  • Development of an algorithm for the transformation of the structure of an input non-stationary information signal through a separate nonlinear dynamic neuron or artificial dynamic neural network and determination of criteria for network parameters in which there are resonance effects in a nonlinear dynamic neuron.
  • Development of dynamic neural network architecture and nonlinear dynamic neuron parameter learning algorithm.
  • Development of an algorithm for recognizing multispectral non-stationary signals based on information resonance effects in a nonlinear dynamic neuron.
  • A computer experiment on recognition of input multispectral images using a three-layer non-linear neural network with dynamic neurons and one hidden layer.

University of Information Technology and Management in Rzeszow

ul. Sucharskiego 2,
35-225 Rzeszow, Poland

phone: 17 866 11 11
fax: +48 17 866 12 22
e-mail: wsiz@wsiz.rzeszow.pl

All rights reserved, University of Information Technology and Management in Rzeszow © 2008-2020