I am a Ph.D. student at the Technical University of Munich. My research focuses on the development of autonomous agents for microeconomic models via deep learning methods. Besides that, I am enthusiastic about any data-related application and uncovering valuable insights.
Recent highlights and an overview of past projects
At the 40th ICML in Honolulu, I had the pleasure of presenting our research paper on an efficient gradient estimation technique for learning optimal strategies in markets.
July 2023
During this workshop in Heilbronn, Stefan Heidekrüger and I presented our research results on equilibrium learning in markets.
September 2022
Several workshops and plenary sessions discussed different aspects of the impact of digitalization on modern societies during this event called upgrade:digitalpolitik. As one aspect of this, I organized a workshop on AI and algorithms and discussed possible regulatory measures, such as the EU AI Act, with the participants.
September 2022
We have released our framework for computing market equilibria via methods of multi-agent deep reinforcement learning called BNELearn.
June 2022
Our research on equilibrium computation has been published in the Nature Machine Intelligence journal. Check out the university's press release or this tweet.
August 2021
The difficulty of predicting developments in the financial markets is to build a machine learning model which can adapt as quickly as possible while not neglecting learned dependencies. The project focused on online deep learning and a random forest modeling approach.
Aug 2019
This work analyzed and modified multiple deep learning approaches for solving regression problems while incorporating uncertainty.
Apr 2019
In this two week class, multiple aspects and the interdisciplinary nature of data science were studied in a cooperation of the University of Tokyo and the University of California, Berkeley.
Jul 2018
In this work, a broad introduction to the machine learning framework TensorFlow by Google was given, and the usage was demonstrated on one real-world application.
Jun 2018
Dropout is a type of regularization in deep learning systems for reducing overfitting by preventing over-complex co-adaptations of neurons on training data.
Jul 2017
The Parareal Algorithm is used for solving initial-value differential equations parallelized over multiple threads. Parareal Algorithm on Wikipedia.
Aug 2016
In this talk, an inner-outer method for solving a large and sparse linear system of equations, such as those occurring in a search-engine problem, is presented.
Oct 2015
List of scientific publications
N. Kohring, F. R. Pieroth, M. Bichler. Enabling first-order gradient-based learning for equilibrium computation in markets. International Conference on Machine Learning, PMLR, 2023.
M. Bichler, N. Kohring, S. Heidekrüger. Learning equilibria in asymmetric auction games. INFORMS Journal on Computing, 2023.
M. Bichler, N. Kohring, M. Oberlechner, F. Pieroth. Learning equilibrium in bilateral bargaining games. European Journal of Operational Research, 2022.
M. Bichler, M. Fichtl, S. Heidekrüger, N. Kohring, and P. Sutterer. Learning equilibria in symmetric auction games using artificial neural networks. Nature Machine Intelligence, 2021.
Short summary of teaching at university
Stylized assessment of programming experience
Feel free to check out more information or contact me on the following platforms