EMCED - Exchange Methods for Contrained Experimental Design (LIT-Project)

For any data-based decision, be it in social sciences and economics, industry or technology, be it in the physical or the virtual world, it is indispensable that the process of collecting the data is as effective as possible. Optimal design of experiments seeks to maximize the information gained from data collection schemes such as eg. monitoring, manufacturing or computer simulation. The latter is a growing field with important applications for automatically generated decisions like in artificial intelligence. In the proposed project we will focus on this area with the specific challenge that experimental design in practice often has to adhere to external constraints. We intend to develop novel algorithms for constructing such designs.