I am a PhD student at the Max Planck Institute for Informatics supervised by Prof. Bernt Schiele and Prof. Matthias Hein . My research interests lie in computer vision and deep learning. Specifically, I am interested in adversarial examples — imperceptibly perturbed images fooling deep neural networks; by understanding the phenomenon of their existence, I want to make deep learning more reliable.
Previously, I obtained both a bachelor and a master degree from RWTH Aachen University . For my master thesis, I worked on weakly-supervised 3D shape completion under the supervision of Prof. Andreas Geiger from the Max Planck Institute for Intelligent Systems and received the STEM-Award IT 2018 as well as the Springorum-Denkmünze . As part of my master degree, I also had the opportunity to spend a semester at Georgia Tech , funded by a scholarship from the Hans Hermann Voss Foundation , working with Prof. Irfan Essa on video segmentation. For my bachelor thesis, I worked on superpixel segmentation under the supervision of Prof. Bastian Leibe .
Over the last few years, I worked for RS Computer , Fraunhofer FKIE , the Computer Vision Group and MATHCCES at RWTH Aachen University , Fyusion , MOBIS and Microsoft . Occasionally, I still do consulting and web development.
On this blog you will find articles, reading notes and some projects — which can also be found on GitHub or ShortScience . Here's my CV , some mission statements , as well as LinkedIn , Xing , and Google Scholar profiles.
What I've been up to ...
In September, I was honored to receive the MINT-Award IT 2018, sponsored by ZF and audimax, for my master thesis on weakly-supervised shape completion. For CVPR 2019, however, I am working on a different topic: adversarial robustness and generalization of deep neural networks.18thOCTOBER2018
In the last few months, I finished my work on weakly-supervised 3D shape completion with a CVPR 2018 paper as well as a follow-up journal submission. This means that I will be visiting CVPR this year. Afterwards, I plan to focus on robustness of deep neural networks — for example: why do adversarial examples exist and how can we defend deep neural networks against them? To answer these questions, I will also be visiting MLSS 2018 later this summer.22thMAY2018
After submitting my master thesis last week, my time at the Max Planck Institute in Tübingen is coming to an end. I will, however, not leave the Max Planck Institute completely. Instead, starting in October, I will start a PhD position at the Max Planck Institute for Informatics in Saarbrücken. Advised by Prof. Bernt Schiele, I will continue research in computer vision and deep learning.09thOCTOBER2017
As part of my master thesis at the Max Planck Institute for Intelligent Systems, I am still trying to use generative deep models to learn how to complete 3D shapes in an unsupervised fashion. As part of Dr. Andreas Geiger's group, I am also working on the KITTI benchmark and was able to visit the Pre-Doc Summer School on Learning Systems in Zürich.10thJULY2017
After starting at the Max Planck Institute for Intelligent Systems in January, I am currently writing my master thesis in Dr. Andreas Geiger's Autonomous Vision Group. In my thesis, I am using generative deep models - including different variants of variational auto-encoders - to learn how to complete 3D shapes in an unsupervised fashion. Beneath my thesis, I am also considering different PhD programs. I applied to the International Max Planck Research School on Intelligent Systems as well as the Max Planck ETH School for Learning Systems. In April, I also had the opportunity to visit Prof. Bernt Schiele's department at the Max Planck Institute for Informatics in Saarbrücken. Interesting times …02ndMAY2017
In the beginning of January, I moved to Tübingen to start writing my master thesis advised by Dr. Andreas Geiger at the Max Planck Institute for Intelligent Systems in the Autonomous Vision Group. Currently, I am getting started with Tensorflow in order to implement new operations for deep learning in 3D. I might share some of my insights regarding Tensorflow in a few articles. Furthermore, I am still thinking about PhD programs in computer vision and autonomous driving starting after my master thesis.14thJANUARY2017
In October, I finally came back home. In the three months at Microsoft, I learned so much about software engineering and project management. But what is more, I learned the importance of self-organization and self-reflection in order to steadily learn and grow - as Microsoft puts it. Now, I am preparing for my master thesis at the Max Planck Institute in Tübingen. This means reading textbooks and papers, taking notes and a bit of brainstorming - check out my reading list. I am also looking for interesting PhD programs starting next year and focussing on computer vision and deep learning for autonomous driving.02ndDECEMBER2016
During the last few months, I deepened my knowledge about deep learning and learned about important computer vision applications for autonomous driving and driver assistance systems. Furthermore, I got an overview about companies as well as state-of-the-art in driver assistance systems. However, after an amazing internship at MOBIS in Frankfurt, it was time to move on to work for Microsoft in Dublin. In the first few weeks, I learned a lot about good software engineering principles - especially in larger teams and projects - and experienced a great working atmosphere.14thSEPTEMBER2016
Before writing my master thesis at the end of this year, I am going to complete two internships - one with Hyundai MOBIS in Frankfurt, Germany and one with Microsoft in Dublin, Ireland. Hyundai MOBIS is a South Korean Tier 1 automotive supplier developing driver assistance systems based on current compute vision research. At Microsoft, I will be working as Software Engineer.05thAPRIL2016
This semester - the last semester before writing my master thesis in summer 2016 - there are several interesting lectures offered at RWTH Aachen University: Advanced Machine Learning by Prof. Leibe, Foundations of Data Science by Prof. Grohe, and Variational Methods in Image Processing (german lecture) by Prof. Berkels. Furthermore, Prof. Seidl offers a lab course on Data Mining on Huge Sensor and Social Network Data which includes the participation in a Kaggle competition. Beneath my studies, I am still working as student research assistant at the Computer Vision Group of RWTH Aachen University.03rdNOVEMBER2015