Research conducted at the Department of Computeer Science
Research conducted at the Department of Computeer Science is placed in the priority area of information technologies and concerns five main directions:
- parallel and distributed processing, clusters and cloud technologies,
- computer methods in economics,
- advanced multimedia biometric techniques and machine learning,
- design and construction of engineering simulation systems,
- optimization and decision support.
In the field of parallel and distributed computing, conducting research enables efficient use of cluster and grid technologies as well as computational clouds. Applications include advanced computing portals, virtual laboratories, interactive simulation systems, etc. Another field of research is the development of methods for efficient use of advanced multi-core processors and hybrid architectures that enable the increase of computing power and optimization of energy demand. An example of these trends are GPU graphics processors and INTEL Xeon Phi accelerators - multi-core architectures that allow significant increase in parallel code performance. Research is conducted on optical networks, in particular flexible optical networks and their integration with wireless access networks. The research focuses on the protocol layer with emphasis on traffic engineering and efficient and optimal graph algorithms applicable to routing and protection of optical connections.
Another field of research concerns modeling and multi-criteria optimization of processes in economics, as well as the application of statistical methods and fuzzy set theory to solve problems of computer-aided decision making. Research is being carried out, among others in the field of optimizing investments in conditions of uncertainty and solving logistic problems. Specialized software is being developed to solve the above-mentioned issues.
In the case of advanced multimedia, biometric techniques and machine learning, the research focuses primarily on the methods of automatic analysis, processing and appropriate coding of digital audio and video signals, in terms of their use in biometric systems, control systems, locating systems, counting and classifying objects. The conducted research focuses on the latest biometric solutions (including the use of various light spectrum spectra), on systems analyzing satellite images, on brain pulse control systems, on bioengineering and robotics components. An important aspect of image recognition is the use of the latest algorithms in the field of neural networks, including deep learning. The entire research is focused around our autonomous electric car.
Another field of research includes the development of computational modules for numerical simulation of physical phenomena and technological processes. The software uses the adaptive finite element method and is built using an object-oriented approach. Implemented software package for the parallel implementation of the finite element method in a model with distributed memory (with message exchange), as well as in a hybrid model, combining models with message exchange and multithreaded. Currently, the software is adapted to the GPU environments.
Research activity on optimization and decision support includes: expert fuzzy systems in asynchronous networks - management, knowledge bases and distribution. In addition, the research is the effectiveness of using genetic algorithms and neural structures in decision support and multicriterial optimization based on uncertain knowledge and multi-criteria optimization in networks.