Full Professor, University of Notre Dame
2 papers at NeurIPS 2025
We treat substructures as graph tokens to enable the scale transformer pre-training on graphs.