Poster
in
Workshop: OPT 2022: Optimization for Machine Learning
Sufficient conditions for non-asymptotic convergence of Riemannian optimization methods
Vishwak Srinivasan · Ashia Wilson
Motivated by energy based analyses for descent methods in the Euclidean setting, we investigate a generalisation of such analyses for descent methods over Riemannian manifolds. In doing so, we find that it is possible to derive curvature-free guarantees for such descent methods. This also enables us to give the first known guarantees for a Riemannian cubic-regularised Newton algorithm over g-convex functions, which extends the guarantees by Agarwal et al for an adaptive Riemannian cubic-regularised Newton algorithm over general non-convex functions. This analysis motivates us to study acceleration of Riemannian gradient descent in the g-convex setting, and we improve on an existing result by Alimisis et al, albeit with a curvature-dependent rate. Finally, extending the analysis by Ahn and Sra, we attempt to provide some sufficient conditions for the acceleration of Riemannian descent methods in the strongly geodesically convex setting.