Tangent Bundle Convolutional Learning: from Manifolds to Cellular Sheaves and Back
This work addresses the problem of learning on manifold-structured data for applications in fields like geometry and machine learning, representing an incremental advancement by connecting to existing sheaf neural networks.
The authors introduced tangent bundle neural networks (TNNs), a novel continuous architecture for processing vector fields on Riemannian manifolds using a convolution operation based on the Connection Laplacian, and demonstrated their effectiveness through numerical evaluation on synthetic and real data tasks.
In this work we introduce a convolution operation over the tangent bundle of Riemann manifolds in terms of exponentials of the Connection Laplacian operator. We define tangent bundle filters and tangent bundle neural networks (TNNs) based on this convolution operation, which are novel continuous architectures operating on tangent bundle signals, i.e. vector fields over the manifolds. Tangent bundle filters admit a spectral representation that generalizes the ones of scalar manifold filters, graph filters and standard convolutional filters in continuous time. We then introduce a discretization procedure, both in the space and time domains, to make TNNs implementable, showing that their discrete counterpart is a novel principled variant of the very recently introduced sheaf neural networks. We formally prove that this discretized architecture converges to the underlying continuous TNN. Finally, we numerically evaluate the effectiveness of the proposed architecture on various learning tasks, both on synthetic and real data.