Berlin Machine Learning & Robotics Lab PI: Marc Toussaint FU Berlin Arnimallee 7 14195 Berlin, Germany marc.toussaint@fu-berlin.de tel: +49 30 838 52485

members

Marc Toussaint

Nikolay Jetchev

Tobias Lang

Nils Plath

Stanimir Dragiev

The MLR Lab focuses its research on the intersection between the fields of Machine Learning and robotics. A specific aspect is the problem of learning appropriate representations to formulate and learn models of the environment on an abstract level. Such compact and behavior-based representations are a precondition for reasoning efficiently about possible behaviors or solutions to problems in natural environments. Eventually, our goal also is to understand how the human brain is able to form such internal representations of the world. The methods we focus on are derived from Reinforcement Learning, latent variable probabilistic models, and probabilistic inference methods.

The group is headed by Prof. Marc Toussaint.

Collaborations

• Honda Research Institute Europe (Dr. Christian Goerick)
• Bielefeld Research Institute for Cognition and Robotics (CoR-Lab) (Prof. Dr. Helge Ritter)
• researchers around the world, e.g.: Pascal Poupart (U Waterloo, work on POMDPs), Nikos Vlassis (TU Crete, work on model-free RL), Matthew Botvinick (U Princeton, work on combining Machine Learning with cognitive science research), Sethu Vijayakumar (U Edinburgh, work on motor control), David Schlangen (U Postdam, work on Bayesian nets in speech recognition systems)

Open Positions

Funding

The group is mainly funded by the Emmy Noether excellence programme of the German Research Foundation under the title of Machine Learning and internal representations in behaviour planning, motor control, and robotics.

Stanio is funded by the collaboration with the Honda Research Institute Europe (Offenbach) and the Research Institute for Cognition and Robotics (CoR-Lab) (Bielefeld).