Ultrahyperbolic Representation Learning
This work addresses the need for more flexible geometric representations in machine learning, particularly for graph data, but it appears incremental as it builds on existing non-Euclidean approaches.
The authors tackled the problem of representing data on non-Euclidean manifolds by proposing a representation on a pseudo-Riemannian manifold with constant nonzero curvature, generalizing hyperbolic and spherical geometries, and applied it to graph representations, providing closed-form distance expressions and gradient-based optimization tools.
In machine learning, data is usually represented in a (flat) Euclidean space where distances between points are along straight lines. Researchers have recently considered more exotic (non-Euclidean) Riemannian manifolds such as hyperbolic space which is well suited for tree-like data. In this paper, we propose a representation living on a pseudo-Riemannian manifold of constant nonzero curvature. It is a generalization of hyperbolic and spherical geometries where the nondegenerate metric tensor need not be positive definite. We provide the necessary learning tools in this geometry and extend gradient-based optimization techniques. More specifically, we provide closed-form expressions for distances via geodesics and define a descent direction to minimize some objective function. Our novel framework is applied to graph representations.