Pellizzoni, Paolo, Till Schulz et al., ‘Graph Neural Networks Can (Often) Count Substructures’, International Conference on Learning Representations 2025 (2025), pp 96013–96039.
journal article · https://proceedings.iclr.cc/paper_files/paper/2025/hash/ef72fa6579401ffff9da246a5014f055-Abstract-Conference.html?utm_source=chatgpt.com