Research Area: Industrial Applications of Machine Learning Techniques

Machine learning techniques are intended to improve data analysis and process engineering. Therefore it is necessary to validate algorithms and approaches not only on artificial data, but also in practical applications, since real tasks have different demands than artificial benchmarks. Our group has successfully carried out several projects with industrial partners in the domain of neural computation and data analysis, as well as reinforcement learning.

• Neural prediction systems for financial time series (1995-1999 at Univ. of Karlsruhe, cooperation with the Helaba)
• Development of an outlier detection system for databases of nuclear contamination (2000-2001, cooperation with the German Federal Research Center for Nutrition)
• Development of a sales rate prediction system for newspapers (since 1996, cooperation with the Axel-Springer-Verlag)

• Optimization of automotive engines (1996 - 1997)
• Optimization of thermostat control (1997, 2002)

See also our publications page for further information.