Original concept of a language-oriented fuzzy regulator
We have developed special software, the LFL Controller (Linguistic Fuzzy Logic Controller). It includes effective algorithms which make the user think that the computer understands normal human speech. The software was used, among other things, for the fuzzy regulation of 5 smelters used for aluminium melting by Al Invest in Břidličná.
The main motivation within this research theme is to create and visualise 3D models of spatial objects with the use of data collected through the 3D reconstruction of image data, CT/NMR volume data and other types of data. The new procedures can be applied in various industries that use computers, from robotic systems and modelling of human tissues for medical purposes, to computer games.
3D modelling of spatial objects
Audiovisual data analysing and speech processing
The objective is to design and implement a system for complex characterisation of individuals and objects through the use of content collected from audiovisual data available in modern communication channels, and test it with real data. The results may be applied, for example, when recognising speech from spontaneous communication between people, between humans and machines, or when identifying a speaker through user authentication processes.
The objective of this theme is to design a set of algorithms for computer vision, mainly through machine learning algorithms (including an evaluation of results). Following from such algorithms, it is possible to create, for example, applications designed for video monitoring, detection of individuals or objects and movement detection.
Computer vision and recognition
Presentation and visualisation
The main objective within this theme is to develop methods for presenting and visualising complex and not very structured data. These procedures will be suitable for personal (single-user) interfaces, including presentations on mobile devices, as well as for interfaces used in multiple-user environments, which include systems with augmented reality.
Mining knowledge from documents and multimedia data means a set of procedures that enable one to obtain knowledge from not very deep-structured electronic documents, from the Internet, metadata extracted from multimedia data - mainly video, or, for example, from the data in molecular biology. These procedures may be applied e.g. in the field of safety and security.
Mining knowledge from documents and multimedia data
Formal languages and grammars
Within the research being conducted in new areas of information technologies, it is necessary to pay more attention to theoretical bases and create new technological principles, for example, in the field of processing augmented information or the methodology for data processing that uses interacting, communicating, distributed and parallel computing environments. Such new knowledge about theoretical bases may result in better new applications of information technologies.
The objective of this theme is to examine the possible use of selected algorithms (for example, simulation, finite element and particle system methods, algorithms for recognising, searching or simulation) used with programmable hardware, GP-GPU, or new architectures, such modern many-core systems or Intel MIC. Another objective is to provide general support for accelerating computing procedures throughout the entire research programme to increase the range of their application.
Acceleration in the field of specialised hardware
Semantic web technologies
Another research theme is the research and development of technologies that may help to implement semantic web concepts. The key element is the effort to simplify the process of creating necessary metadata and make it more effective and efficient. Possible applications are seen mainly in web development, thus providing and getting more information through the web.