Molecular property prediction, QSAR/QSPR, and applications: molecular fingerprints, physicochemical descriptors, and pretrained molecular foundation models. We apply them to concrete problems and domains like ecotoxicology and peptide function prediction.
Graph theory and learning on graphs: topological descriptors, graph neural networks, graph classification. We research structural baselines and graph-theoretic aspects.
Constructing benchmarks and datasets for machine learning and chemoinformatics. We focus on fair evaluation, strong baselines, and statistical analysis methods.
Predictive modeling and novel artificial intelligence applications, e.g. pen & paper RPGs.
Software for chemoinformatics, machine learning, and data processing. We provide high-quality libraries and reproducible pipelines.