Enterprise References

Drug Discovery

Partners:

• ComGenex Inc.
• ComGrid Ltd.

One of the basic issues of modern drug discovery is the exclusion of chemically unstable, biologically inactive and toxic compounds from the research process in the early stages, thereby reducing the cost and the time period of the drug development. The main purpose of the ADMEToxGrid project is to develop an enterprise Grid system that is suitable for predicting these chemical parameters of millions of compounds in a short time and in a secure manner, while also exploiting the free capacity of the office computers located at the different site of the company. In this project SZTAKI Desktop Grid serves as the base of the Grid framework, and it will be extended with new features such as advanced interface to databases.

More information: www.ademtoxgrid.hu

Datamining and machine learning

Partners:

• MTA SZTAKI Machine Learning and Human-Computer Interfaces Group
• University of Szeged, Artificial Intelligence group
• AAM Consulting
• T-Systems Hungary

The project aims the development of the prototype in a datamining software using SZTAKI Desktop Grid technology. The software enables the user to select the algorithm and to make scheduling decisions, as well as the generation of higher quality datamining models by automating these permits. The innovative element of the project is the optimization in the scheduling of datamining algorithms enabled by meta-level learning. The prototype supports the documentation and verification of datamining projects, while it remains expandable thanks to its architecture. Special attention is paid to data privacy issues. After the termination of the project, the prototype and its subsequent versions will be available for non-profit research purposes. Following up on the results of the project, the members will consider the commercial deployment of a datamining grid-based product.

More information: www.sztaki.hu/search/projects

Academic References

Information Theory

Partner:

• ELTE, Department of Computer Algebra

The aim of this project is to find all the generalized binary number systems up to dimension 12. In the application, an extensive search is performed in the finite set of matrices of given size fulfilling some necessary conditions. The difficulty is that the size of this finite set is an exponential function of the dimension. The result of the application is a list of characteristic polynomials that are already likely to be number system bases. Later on, this list is post-processed, and the final result is a complete list of binary number systems in a fixed dimension. Based on the results, ELTE performs information theoretical analysis. The result of this is a list of number systems that could be successfully applied in data compression, coding or cryptography. Moreover, number systems are interesting in a geometrical aspect of view. E.g.: topological analysis, calculation of the dimension, connectedness etc. Finally, knowing all matrices up to a given dimension could help us to a deeper understanding of the mathematics of generalized number systems.

The peak performance of the SZTAKI Desktop Grid based public project is over 1 TFlop/s with more than 12.000 CPUs. Thus, in 2006/Q1 the established system became one of the fastest computational infrastructures operated from Hungary.