Collaborative Research Center SFB 876
Providing Information by Resource-Constrained Data Analysis
Project description
DFG has established eleven new Collaborative Research Centres, two of which the University of Duisburg-Essen participates in with funding worth 585.400 Euro.
The new Collaborative Research Centre 876, "Providing Information by Resource-Constrained Data Analysis", is coordinated through the Technical University of Dortmund. It focues on different aspects of data processubg - from the analysis of large data sets and energy efficiency of small equipment to the intelligent connection of sensor data. It aims at gaining information from ever increasing data amout; in a timely manner, without high energy demand and on location. Along with TU Dormund, the Leibniz Institute for Analytical Science (ISAS) and the University of Duisburg-Essen are involved in the project.
In UDE project "Analysis and communication for dynamic traffic prediction" explores novel, high-precision real-time prediction methods for the behavior of object streams, such as vehicles, goods or people. Project leader Prof. Dr. Michael Schreckenberg: "The quality of the forecasts depends on several factors. Among other things it depends on how extensive and accurate the data is."
Given the availability of new sensor data sources, the subproject explores new methods for data compression and evaluates the achievable improvements of the forecasts. The plan is to transfer these results to other areas later in order to optimise logistical processes. Another partner in this subproject is Prof. Dr.-Ing. Christian Wietfeld of the TU Dortmund.
The second UDE subproject is concerned with the large amounts of data, arising in molecular biological investigations ("Feature selection in high-dimensional data using the example of the risk prediction in oncology"). The project is lead by PD Dr. Alexander Schramm from the oncology laboratory of the Children's Hospital. In cooperation with Prof. Katharina Morik of TU Dortmund, information relevant to cancer patients will be filtered from the datasets. Prof. Schramm: "This is an important prerequisite for better individual therapies and for the exclusion of unnecessary treatments. Due to their molecular profile they can in fact not even be used on some patients." In the long term, actionable clinical test methods will be derived from this research.