Hi, my name is Claas Voelcker. I am a doctoral student in computer science focussing on the theory of reinforcement learning and machine learning. I am employed at the University of Technology in Darmstadt under the supervision of Kristian Kersting and Jan Peters. I will begin a new post at the University of Toronto and the Vector Institute under the supervision of Amir-massoud Farahmand and Animesh Garg. In addition, I am an active member of Queer in AI and local queer activism.
My main research interests lie at the intersection of reinforcement learning, probabilistic inference and representation learning. My overarching goal is to get a better understanding of when algorithms fail and what enables specific algorithms to tackle certain kind of environments beyond the vague intuition that is often prevalent in current machine learning research. To achieve this, I believe we need a more well founded theory of what specific aspects of MDPs agents actually (ab)use to achieve their goals.
Besides conducting research, I am a connoisseur of failed cooking experiments. In the past I have been accused of having a very stern resting face, of being political to a fault and of being way to excited in the morning. As for my virtues, people frequently list: not failing at cooking experiments as often as I fear, not being a complete bore and having a pretty decent singing voice.
According to my classmates at university, I am mad, completely insane or just plain wrong, because I actually like math and statistics.