Research interests

My name is Astrid Klipfel and I really like using theoretical frameworks like group theory, graphs and circuits to solve real-world applications.

I did my PhD thesis entitled Equivariant Message-Passing Neural Networks and Probabilistic Diffusion Models for Materials Science at the University of Artois from 2020 to 2023. My research was on graph neural network architectures and generative models for materials science. During this period, I proposed a method to build MPNN equivariant to both $$\text{E}(3)$$ and the repaving group $$\text{SL}_3(\mathbb{Z})$$ (E-MPNN), an efficient equivariant MPNN using vector fields (Vector fields MPNN) and a diffusion model for periodic materials (Diffusion Model).

Currently I'm a postdoctoral researcher at the Faculty of Electrical Engineering (CTU) in Prague and I'm very interested in Neuro-symbolic AI, first-order logic and probabilistic circuits.