Teaching
- MATLAB Course
Tutorials, Winter semesters 2019/2020, 2020/2021, 2021/2022
Led tutorial sessions to answer questions from the participants, prepared and graded assignments on MATLAB basics such as plotting, looping, efficient vectorization, and implementation of optimization algorithms.
- Computational Methods of Environmental Modeling
Exam support, Winter semesters 2018/2019, 2019/2020, 2020/2021
Prepared and graded exams on topics such as linear equations, numerical methods (to solve nonlinear equations, integration, and ordinary differential equations), as well as uncertainty quantification.
- Environmental Fluid Mechanics
Exam support, Summer semesters 2018, 2019, 2020, 2021, 2022
Prepared and graded exams on topics such as advection-diffusion equation, fluid flow in porous media, and relevance of processes in fluid flow.
Supervision
- Hybrid modeling of shock waves in hydraulic lines with neural delay differential equations
Peter Phan Dinh Nguyen, 2022 M.Sc. thesis
This thesis was commissioned by Bosch, specifically the group of Dipl.-Ing. Alexander Fischer and Dr.-Ing. Andreas Klein, to develop a neural delay differential equation model that is able to capture nonlinear dynamics (shock waves, state-dependent fluid properties), for accurate predictions of pressure and velocity at the ends of a hydraulic line.
- Proposing analytical scientific equations with sparse symbolic-layered neural networks
Larissa Brencher, 2022, M.Sc. thesis
This thesis investigated the equation learner model to learn symbolic representations as a differentiable counterpart of the symbolic regression method.
This thesis was intended to explore the possibility of integrating the equation learner model in my FINN model as a replacement for the neural network modules (read more) to further improve the model interpretability.
- Using Machine Learning Tools to Improve a Conceptual Hydrological Model of the Upper Neckar Catchment
Kamyar Chabok, 2022, M.Sc. thesis
This thesis explored various ways to improve the predictions of LSTM models by feature selection and engineering, as well as combining information obtained from an HBV model.
- Markov Chain Monte Carlo for Artificial Neural Networks
Andrei Chalapco, 2021, B.Sc. thesis
This thesis investigated the pre-conditioned Crank Nicolson MCMC (pCN-MCMC) as a relatively cheap MCMC algorithm to be applied on Bayesian Neural Networks.
This thesis was intended to explore the feasibility of an MCMC algorithm to be integrated in my model FINN to quantify its parameter and prediction uncertainty (read more).
- Prediction of water levels in a wetland-groundwater system influenced by drinking water extraction using an artificial neural network
Sascha Flaig, 2020, M.Sc. thesis
This thesis was commissioned by the engineering firm Kobus und Partner, specifically co-supervised by Dr.-Ing. Ulrich Lang and Dr.-Ing. Alexander Kissinger.
This thesis developed an LSTM model to predict the swamp water level, affected by the nearby well pumping for drinking water production.
This thesis resulted in an EGU conference talk (read more).
- Machine Learning for Thermochemical Energy Storage Device
Thilo Walser, 2019, M.Sc. thesis
This thesis developed a physics-informed NARX model to predict the internal states of a thermochemical energy storage device and to ensure the physical consistency and plausibility of its predictions.
This thesis resulted in a paper (read more).
- Energy capacity of a geothermal reservoir - The effect of conductive heat transfer recharge on reservoir lifetime at low temperature conduction regimes
Arjan Marelis, 2017, B.Sc. thesis
This thesis investigated the impact of including the surrounding rock formation on a geothermal reservoir simulation as a source of heat recharge.
This thesis used the geothermal reservoir simulator that I wrote for my master's thesis.