3 PhD Scholarships in Machine Learning Metamodeling Architectures for Scientific Simulation – DTU Management at Technical University of Denmark

The goal of these 3 PhDs is to develop different types of machine learning-based approximations to simulation tools, that can accelerate their runtimes by many orders of magnitude. If this happens with good accuracy in- and out-of-distribution, then we can eventually apply them directly in such policy studies.

The 3 PhDs are collaborative yet independent, because they explore methodological concepts in three groups:

Active Learning and Causal Discovery
Graph Neural Networks and Neural Architecture Search
Polynomial Neural Networks and Physics-informed deep learning