Scalable and equivariant spherical CNNs by discrete-continuous (DISCO) convolutions

Publication
International Conference on Learning Representations (ICLR)