Stochastic Weight Matrix-based Regularization Methods for Deep Neural Networks
Reizinger P., Gyires-Tóth B. (2019) "Stochastic Weight Matrix-Based Regularization Methods for Deep Neural Networks.." Springer LNCS 1. 11943.
CV
Short biography
Patrik Reizinger is a PhD student at the University of Tübingen, supervised by Wieland Brendel, Ferenc Huszár, Matthias Bethge, and Bernhard Schölkopf. He is part of the ELLIS and IMPRS-IS programs. His main research interests include causal inference and representation learning. He has done both his MSc and BSc at the Budapest University of Technology in electrical engineering and specialized in control engineering and intelligent systems.
Education
PhD in Machine Learning, University of Tübingen
2021-2024 (expected)
- Affiliations: IMPRS-IS, ELLIS
- Research interests: causal representation learning, Independent Component Analysis, identifiability
- Supervisors: Wieland Brendel, Ferenc Huszár, Matthias Bethge, and Bernhard Schölkopf
MSc in Electrical Engineering, Budapest University of Technology and Economics
2019-2021
- GPA: 5.0/5.0
- Thesis: Development of an Attitude Determination and Control System for CubeSats on LEO orbits
- Supervisors: Ferenc Vajda, Márton Szemenyei
BSc in Electrical Engineering, Budapest University of Technology and Economics
2015-2019
- GPA: 5.0/5.0
- Thesis: Development of a 3D input device for virtual working environments
- Supervisors: Ferenc Vajda, Márton Szemenyei
- Exchange semester: Karlsruhe Institute of Technology, Germany
- Affiliations: German language program, Integrated MSc-BSc program (IMSc)
Work experience
C3S Electronics LLC
February 2019-March 2021
- Role: Research Engineer
- Project: Control and estimation system develeopment for CubeSats (ADCS)
- Supervisor: Ferenc Vajda
- Topics: Kalman Filtering, Hybrid control
- Technologies: Python, C++14/17, Catch2, Confluence
Karlsruhe Institut for Technology
Winter 2019
- Role: Research Assistant
- Project: Time synchronisation in FPGAs
- Supervisor: Vladimir Sidorenko
- Technologies: Python
Budapest University of Technology and Economics
September 2016-January 2021
- Role: Research Assistant
- Project: Multi-agent reinforcement learning, Generalization in deep learning, Embedded systems
- Supervisors: Bálint Gyires-Tóth, Márton Szemenyei, Ferenc Vajda
Fraunhofer Institute for Factory Operation and Automation IFF
Summer 2018
- Role: Image Processing Intern
- Project: Automated visual inspection tool development
- Supervisor: Thomas Dunker
- Technologies: C++11/14, Python
Gravity R&D LLC
Summer 2017
- Role: Data Scientist Intern
- Project: Customer behavior analysis
- Supervisor: Bottyán Németh
- Technologies: Python
Skills
Programming
- Python
- C++11/14
- PyTorch
- Weights and Biases
- Git, GitHub
Research Management
- Confluence
- Jira
- Notion
- Overleaf/LaTeX
- Zotero
- ResearchRabbit
Publications
Attention-based Curiosity in Multi-agent Reinforcement Learning Environments
M. Szemenyei and P. Reizinger. (2019). "Attention-Based Curiosity in Multi-Agent Reinforcement Learning Environments" International Conference on Control, Artificial Intelligence, Robotics & Optimization (ICCAIRO).
Attention-based Curiosity-driven Exploration in Deep Reinforcement Learning
P. Reizinger and M. Szemenyei. (2020). "Attention-Based Curiosity-Driven Exploration in Deep Reinforcement Learning" 2020 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP).
Learning to Play Robot Soccer from Partial Observations
M. Szemenyei and P. Reizinger. (2020). "Learning to Play Robot Soccer from Partial Observations" 23rd International Symposium on Measurement and Control in Robotics (ISMCR).
Talks
Teaching
Service and leadership
- Mathias Corvinus Collegium Leadership Academy
- Formula Student East EV/DV Judge