Fabian Jogl co-authored the paper “Maximally Expressive Graph Neural Networks for Outerplanar Graphs,” published in the Transactions on Machine Learning Research (TMLR). The work proposes a linear time graph transformation enabling the Weisfeiler-Leman algorithm

and message passing graph neural networks to achieve maximum expressivity on outerplanar graphs — a graph family covering most small pharmaceutical molecules. By encoding the Hamiltonian cycle of each biconnected component in linear time, the method achieves

provably maximum expressivity on this graph family at negligible computational overhead, with experiments confirming improved predictive performance on molecular benchmarks.