Machine Learning Control – AMB-ML

Machine Learning Control (MLC) is a technique that deeply interconnects physics and computer science. Particularly, the use of heuristics to control non-linear processes requires very novel methods and is the subject of the Turbulence Control Group of Pr. Bernd R. Noack.

A novel approach to MLC is to use a model free strategy applying Genetic Programming (GP) to find non-linear control laws. The idea was born in January 2013 during the first International Forum for Turbulence and Network Control, when Pr. Noack, Pr. Abel and Dr. Segond were discussing about the implications of artificial intelligence in physics. This approach is being, since then, developed by Ambrosys GmbH in a generic unified framework for MLC (AMB-ML).

AMB-ML is a generic, ready to use, framework that is able to find control rules for any system that is able to provide sensor informations and take actuation information in a standardized form (HTTP communication in JSON format).