Marius Memmel

I am a PhD student at the University of Washington supervised by Prof. Dieter Fox (RSE lab) and Prof. Abhishek Gupta (WEIRD lab).

In 2022, I graduated with a Master's in Computer Science from TU Darmstadt, focusing on computer vision and robot learning, and minoring in Entrepreneurship & Innovation.

I was on an exchange at EPFL, where I wrote my thesis at VILAB supervised by Prof. Amir Zamir (and formally Prof. Stefan Roth). I was a student research assistant at the AI&ML Lab of Prof. Kristian Kersting (2020-2021) and MEC-Lab of Dr. Anirban Mukhopadhyay (2020-2021) and collaborated with the IAS group of Prof. Jan Peters (2020-2021). Furthermore, I worked as a part-time Machine Learning Engineer at Sopra Steria SA (2020).

In 2019, I received my Bachelor of Science in Business Information Systems from the Baden-W├╝rttemberg Cooperative State University Mannheim (DHBW) and got awarded the Best Graduate Award for my academic achievements. As part of the program, I worked part-time as a Software Engineer at Knauf IT (2016-2019) and took part in an exchange with Appalachian State University (2018).

Email  /  CV  /  Google Scholar  /  Twitter  /  Github  /  LinkedIn

profile photo
Research

I'm interested in building visual perception systems that make active agents robust to changes and help them generalize to new environments. I believe the interplay between perceiving the environment and interacting with it is the next step towards creating intelligent, multi-purpose agents. Additional interests include improving exploration in reinforcement learning through intrinsic rewards like, e.g., curiosity. (* equal contribution)

News
  • 06/2022: Our paper "Modality-invariant Visual Odometry for Indoor Navigation" has been accepted to the Embodied AI Workshop at CVPR 2022.
  • 03/2022: Our paper "Interactive Disentanglement: Learning Concepts by Interacting with their Prototype Representations" has been accepted to CVPR 2022.
  • 01/2022: Our paper "Dimensionality Reduction and Prioritized Exploration for Policy Search" has been accepted to AISTATS 2022.
Research

Peer reviewed conference and workshop papers.

Modality-invariant Visual Odometry for Indoor Navigation
Marius Memmel, Amir Zamir
Embodied AI Workshop at Conference on Computer Vision and Pattern Recognition (CVPR), 2022

paper    poster    code   

Interactive Disentanglement: Learning Concepts by Interacting with their Prototype Representations
Wolfgang Stammer, Marius Memmel, Patrick Schramowski, Kristian Kersting
Conference on Computer Vision and Pattern Recognition (CVPR), 2022

paper    arxiv    code   

Dimensionality Reduction and Prioritized Exploration for Policy Search
Marius Memmel, Puze Liu, Davide Tateo, Jan Peters
International Conference on Artificial Intelligence and Statistics (AISTATS), 2022

paper    arxiv    code    poster   
Adversarial Continual Learning for Multi-Domain Hippocampal Segmentation
Marius Memmel, Camila Gonzalez, Anirban Mukhopadhyay
Domain Adaptation and Representation Transfer (DART) at Medical Image Computing and Computer Assisted Intervention (MICCAI), 2021

paper    arxiv    code    poster    video   
Other Projects

Thesis, projects, open-source contributions, and miscellaneous. For my past data science and software engineering projects please refer to my GitHub.

Multi-modal Vision Transformers For Data-Efficient Visual Odometry In Embodied Indoor Navigation
Master Thesis, Technical University of Darmstadt, 2022

Thesis at VILAB (EPFL) supervised by Prof. Amir Zamir and Prof. Stefan Roth.

thesis    code   
Open Source Contribution to MushroomRL
Intelligent Autonomous Systems Group, TU Darmstadt, 2021

Implementation of Constrained Weighted Maximum Likelihood Update for REPS (CREPS) and Model-based Relative Entropy Policy Search (MORE).

code   
DeepFovea++: Reconstruction and Super-Resolution for Natural Foveated Rendered Videos
Artificial Intelligence & Machine Learning Lab, TU Darmstadt, 2020

Group project for the Deep Learning Architectures and Methods course supervised by Prof. Kristian Kersting.

report    code   
Deep Learning for Case-based Reasoning through Prototypes: A Neural Network that Explains Its Predictions PYTORCH IMPLEMENTATION
Artificial Intelligence and Machine Learning Lab, TU Darmstadt, 2020
code   
EmojiGAN: Image to Emoji Translation Using GANs
Medical & Environmental Computing, TU Darmstadt, 2020

Group project for the Deep Generative Models course supervised by Dr. Anirban Mukhopadhyay. Special thanks to Tillmann Rheude (image).

code   
Creative Chest X-Ray Augmentation Using Cycle GAN
Medical & Environmental Computing, TU Darmstadt, 2019

Group project for the Deep Learning for Medical Imaging course supervised by Dr. Anirban Mukhopadhyay.

poster   
Scalable 3D Semantic Segmentation for Gun Detection in CT Scans
Marius Memmel*, Christoph Reich*, Nicolas Wagner*, Faraz Saeedan
Visual Inference Lab, TU Darmstadt, 2021

Group project for Deep Learning for Computer Vision supervised by Prof. Stefan Roth.

arxiv    code   

Conception and Development of a Machine Learning Model for the Analysis of the Energy Consumption of a Plasterboard Dryer
Bachelor Thesis, DHBW, 2019

Thesis at Knauf IT supervised by Dr. Halgurt Bapierre and Prof. Julian Reichwald.

thesis   
Honors & Awards
  • 2022: European Informatics Student Award: "An incentive for the brightest European students to experience the University of Washington and the Pacific Northwest."
  • 2019-2020: Deutschlandstipendium: Merit-based scholarship given to less than 1% of all students in Germany
  • 2019: Best Graduate: Award for the best graduate of Business Information Systems (Software Engineering) at DHBW
  • 2018: Baden-W├╝rttemberg-Stipendium: given to 1500 high-achieving students/year to promote exchange
English Proficiency
  • 07/07/2020 TOEFL iBT: total 112/120, reading 29, listening: 29, writing 28, speaking 26
  • 12/11/2021 Duolingo English Test: total 150/160 (equivalent to 120/120 TOEFL iBT), literacy 150, comprehension 150, conversation 155, production 150

Design and source code from Jon Barron's website.