Search papers, labs, and topics across Lattice.
1
0
1
0
SPD-SheafNets learn richer geometric representations than standard GNNs by operating directly on matrix-valued features, achieving SOTA on molecular property prediction by capturing relationships between directions.
