
The author proposes LaB-GATr, a transformer neural network with geometric tokenization for large biomedical meshes, achieving state-of-the-art results in cardiovascular hemodynamics modeling and neurodevelopmental phenotype prediction.


coremsg

lab-gatr-geometric-algebra-transformers-for-large-biomedical-meshes


LaB-GATr: Geometric Algebra Transformers for Large Biomedical Meshes


title_rewrite


LaB-GATr is a powerful architecture for learning with high-fidelity meshes in biomedical applications.
