Hi, I'm Gregor. My interests include interpretable and explainable artificial intelligence, probabilistic machine learning, efficient deep learning, and sample-efficient model-based reinforcement learning. I believe research in these areas can drive forward the progress and open new possibilities in applications like autonomous driving, robotics, prosthetics, geriatrics and many others.

Reinforcement learning can be seen as being inspired by an intelligent organism acting in an environment and adapting his next actions based on biological reward signals. Model-based reinforcement learning allows for the agent to keep a model of the environment, akin to human imagination, and use it to make better predictions. Deep reinforcement learning has shown remarkable results in various applications in recent years. However, in real-world applications one run can be expensive or even dangerous. Sample-efficient approaches manage to learn quickly from a limited amount of data.

Deep learning has achieved significant results in numerous areas at the cost of larger and larger models which are becoming increasingly impossible for a human to understand. Explainable artificial intelligence can help avoid those black box models. One way of doing this is probabilistic machine learning, other methods include attention based methods.

On this site I present various concepts, tutorials, and projects.