Homepage News

Rzeszów / Researchers developed a more effective "learning" system for artificial intelligence

28.07.2017 Technology, Universities

Photo: Fotolia

Researchers from Rzeszów have developed a new "learning" system for artificial intelligence. It allows to much more effectively analyse information, associate causes with effects and correct errors based on past experiences.

Researchers from the Department of Electronics and Telecommunications at the University of Information Technology and Management in Rzeszów have developed new algorithms for learning systems, which - as PAP said one of the scientists, Dr. Janusz Kolbusz told PAP - give better results and are much more effective than previously known and used algorithms.


The new system fulfils three conditions at the same time: it is faster than the previous systems; it is more efficient (it needs less computing power to do the same job); and gives better results.


"It is very difficult to write algorithms for learning systems. We have been able to create an algorithm that can teach any architecture. We have also developed architectures that are small, but have enormous computing power. We have also created many new methods of neuronal connections, allowing to build up to 100 times stronger networks" - said Dr. Kolbusz.


He added that it was "the only such system in the world right now".


In his interview with PAP, he explained that deep learning, involving the creation of complex neural network architectures, is currently the fastest growing area of artificial intelligence. Thanks to such neural networks, machines are now able to learn, recognize objects and react like humans, and often even better. Such solutions are used, for example, in medical diagnostics, speech recognition, financial market forecasting, and in the near future they may lead to the development of self-driving vehicles or intelligent robots.


So the innovation, modernity and originality of the system developed by the scientists in Rzeszow is determined not only by the fact that artificial intelligence can independently analyse data, draw cause-effect conclusions and correct errors, and consequently solve complex problems, but also that thanks to the developed algorithms it performs these operations more efficiently and with better results.


The new algorithms have been developed by the team led by Prof. Bogdan Wilamowski, composed of: Dr. Janusz Korniak, Dr. Janusz Kolbusz, Dr. Paweł Różycki.


As an example of the use of artificial intelligence, professor Wilamowski mentioned the American stock exchange, where learning systems compete with each other, not people.


"A man is not able to fully understand the stock exchange: there are numerous movements, continuous change. A man may try to analyse it, but can not cover all the factors at once because of biological limitations" - the scientist pointed out.


"If someone plays the stock market and thinks they can get rich, they either have connections to companies and acquire information illegally, or they will certainly lose. A person gets stock market information with a few minutes delay, and neural networks make decisions in microseconds" - he added.


Dominik Łazarz from the press office of the Rzeszów university told PAP that the research of scientists from the University of Information Technology and Management met with interest in the world of science both in Poland and in the world. Their results have been published in prestigious scientific journals such as: IEEE Transactions on Industrial Informatics, IEEE Transactions on Industrial Electronics, IEEE Transactions on Electron Devices, International Journal of Numerical Modelling: Electronic Networks, Devices and Fields.


The research project was also recognized by the Minister of Science and Higher Education as one of the most important scientific achievements and included in the publication "Achievements of Polish Science 2016". The National Science Centre co-financed the project with the amount exceeding PLN 416 thousand.


Agnieszka Pipała (PAP)


api/ agt/ kap/


tr. RL

Share this
Rating: 0 votes


Forgotten password


Comments: 0
Add comment See all comments  

Most popular sites