Feb 20, 2023 · Here, we simplify such difficulties for a class of sparse or structured symmetric positive-definite matrices with the affine-invariant metric.

Our local-parameter approach gives a computationally efficient paradigm for a class of SPD sub- manifolds where it can be nontrivial to design an efficient.

A generalized version of the Riemannian normal coordinates is proposed which dynamically trivializes the problem into a Euclidean unconstrained problem and ...

Feb 21, 2023 · Riemannian submanifold optimization with momentum is computationally challenging because ensuring iterates remain on the submanifold often ...

Jul 23, 2023 · We do so by proposing a generalized version of the Riemannian normal coordinates that dynamically orthonormalizes the metric and locally ...

Riemannian submanifold optimization with momentum is computationally challenging because, to ensure that the iterates remain on the submanifold, we often ...

We use our approach to explain and simplify existing approaches for structured covariances and develop efficient second-order optimizers for deep learning ...

Riemannian submanifold optimization with momentum is computationally challenging because, to ensure that the iterates remain on the submanifold, ...

我们使用仿射不变度量为对称正定矩阵的子流形简化了此类优化算法。我们提出了黎曼法线坐标的广义版本，它动态地将问题平凡化为欧几里得无约束问题。我们使用我们的方法来 ...

Simplifying momentum-based positive-definite submanifold optimization with applications to deep learning. W Lin, V Duruisseaux, M Leok, F Nielsen, ME Khan, M ...