This BMBF-sponsored project was conducted in 2009-2013. Project partner were Leiden University, NL, University Bochum, Nurogames GmbH, Cologne and divis GmbH, Dortmund.
Systematic optimization of models for complex applications in IT and automation, here with the goal of forecasting and optimal control of plants or processes, is the topic of this project. It poses still a great challenge for the practioner in computer science or engineering. In many cases it is not alone a problem of the right model parametrization, but also a task of intelligent data preprocessing and data selection. Here the project SOMA aims at developing and offering new solutions and tools.
Under the umbrella of SOMA several sub-projects were undertaken: Tuned data mining (TDMR), gesture recognition, reinforcement learning for strategic games, intelligent methods for feature generation like slow feature analysis (SFA) or n-tuple systems.
Topics: Applied computer science, modeling, simulation, learning system, computational intelligence (evolutionary algorithms, neural networks), data mining.