Seven renowned scientists from Europe and the US, including Sándor Baran, head of the Department of Applied Mathematics and Probability Theory of the Faculty of Informatics, have become scholars of ECMWF as of 1 January 2021.
The European Centre for Medium-Range Weather Forecasts (ECMWF) is a research institute that also provides global numeric weather forecasts for 34 member countries.
In their first 3-year period, scholars are involved in modelling the weather system of the Earth, ensemble forecasts, and research areas related to the applications. Sándor Baran’s project is focused on the statistical post-processing of ensemble forecasts.
The aim of the Open Clouds for Research Environments (OCRE) project is to present the positive effects of cloud-based and digital services on research findings in terms of agility, and to facilitate the use of cloud technologies in the European research community with the support of service providers and universities.
Launched in November, the OCRE programme is about the use of cloud services available through the OCRE IaaS + framework agreements. Projects that may be case studies for how services can support research were invited. 27 projects were submitted from 12 countries, representing a wide variety of educational areas. 15 applicants were successful in the first round.
One of these was András Hajdu, head of the Department of Computer Graphics and Image Processing with his work entitled “Approximation of theoretical models and application-oriented further evolvement”. The project provides EUR 50,000 for cloud computing capacity for 6 months. Members of the research group are PhD students Gergő Bogacsovics, Róbert Lakatos, and Attila Tiba.
By analysing theoretical models, the scientists involved will develop neural circuit-based solutions that are more accurate than the ones currently used. To this end, they will use Data Science, Big Data, Deep Learning, and for data processing cloud computing services (Microsoft Azure).
Currently, in certain cases it takes a considerable amount of time to develop complex neural circuits, even with supercomputers. For this reason, research teams and companies with smaller computing capacity mostly use pre-formed circuits. In the framework of the OCRE project, using the Microsoft Azure platform and this method researchers and businesses will be enabled to create high-quality pre-formed artificial neural circuits without a huge amount of data.