Scholar
Oleg Arenz
Google Scholar ID: fSyGIDwAAAAJ
Postdoctoral Researcher, Technische Universitaet Darmstadt
Citations & Impact
Citations
356
H-index
11
i10-index
12
Publications
20
Co-authors
3
list available
Publications
Resume (English only)
Academic Achievements
- Developed GMMVI software framework: A high-performance and well-documented Gaussian mixture model optimization framework for variational inference using natural gradient descent. The framework is quite modular, supporting different techniques, such as estimating the natural gradients or selecting the samples for each update. It supports a total of 432 different combinations of design choices.
Research Experience
- During his PhD, he investigated several different learning problems for robotics, including reinforcement learning, inverse reinforcement learning, and variational inference, showing that they can all be framed as an information projection, a particular type of distribution-matching problem. By treating the aforementioned learning problems as different instances of an information projection, they can be solved based on similar insights. For example, an upper bound on the I-Projection objective was derived and used in combination with an expectation-maximization procedure for variational inference, density estimation, as well as non-adversarial imitation learning.
Education
- PhD (2015-2020), Advisor: Gerhard Neumann (currently full professor at Karlsruher Institute of Technology).
Background
- Research Interests: Machine Learning, Robotics, Inverse Reinforcement Learning, Imitation Learning, Grasping and Manipulation, Reinforcement Learning, Variational Inference. Affiliated with TU Darmstadt, Intelligent Autonomous Systems, Computer Science Department.
