Big Data Research
In these days information coming from several sensors is often referred to as one of the ‘Biggest Big Data’ source. A typical example for these data is the stream we receive by the transmission of security video cameras. Exploiting data by image acquisition is a rather simple and relatively cost-effective technique, so it is very popular when the effects of different interventions are checked. It is a common approach when e.g. the reaction of cells/cell cultures to drugs or light are continuously recorded by video cameras. In this field we plan to develop distributed and sufficiently precise algorithms to process online streams. The exploitation is very diverse, so the techniques that are planned to be elaborated to the target field, besides processing videos for different aims, can also be used for sensors which do not transmit image information, but some different type of data. In the case when the observation of the examined object/event is done by several sensors, we plan to develop fusion techniques by the application of ensemble-based systems.
Efficient and scalable visualization of Big Data
Having a comprehensive view of the visualisation of Big Data is a problem in many fields; however, models generally suggested are in most cases far from being optimal. We wish to examine how general visualization models can be developed further according to the users’ needs, e.g. meeting the demand of 3D sensor controllability, effective representation of attributes and relations and the ability of assigning annotations. As different forms of representation are optimal for different data structures, we plan to organize the information deriving from Big Data by data mining in hierarchical structures in which different visualisation models can be assigned to the given levels. Beyond the structural representation, the scalable representation of Big Data is also an important task for which we plan to develop fractal-based models. This sort of representation is typically suitable for the visualisation of genetic data.
