Technische Universität München

Open Opportunities

Hybrid physics-ML modeling with zero-th order optimization Technische Universität München Helmholtz Center Munich In the context of numerical climate and weather modeling, we aim to replace the expensive function “run a high-resolution simulation and then coarse-grain” with a hybrid physics-ML model with an in-the-loop data-driven corrector, learned via a zero-th order optimization method. Artificial Intelligence and Signal and Image Processing, Dynamical Systems, Knowledge Representation and Machine Learning Master Thesis

Optimizing a Minimal Language using Pre-trained Language Models Technische Universität München Helmholtz Center Munich Constructed languages (conlangs) have been a topic of interest in linguistics and language design for centuries. Recent advances in natural language processing (NLP) and machine learning (ML) have opened up new possibilities for conlang design. This research proposes to explore the use of pre-trained language models (LLMs) to develop a new conlang, with a focus on creating an optimal minimal language. Artificial Intelligence and Signal and Image Processing, Knowledge Representation and Machine Learning, Learning, Memory, Cognition and Language Master Thesis

Bayesian Physics-Informed Neural Networks Technische Universität München Helmholtz Center Munich The goal of this project is to combine physics-informed neural networks (PINNs) with Bayesian deep learning (BDL) to develop a novel neural network that is both informed about the underlying data and is capable of uncertainty estimation. Artificial Intelligence and Signal and Image Processing, Knowledge Representation and Machine Learning Master Thesis