Learning algorithms utilizing quasi-geodesic flows on the Stiefel manifold

182 citations recorded by the source corpus.

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

Learngene: From Open-World to Your Learning Task (2021, neutral): “elements of the Stiefel manifold [44] Vd,k.”
Efficient Riemannian Optimization on the Stiefel Manifold via the Cayley Transform (2020, methodology): “Other approaches are based on the closed-form Cayley transform (Fiori et al., 2012; Jiang & Dai, 2015; Nishimori & Akaho, 2005; Zhu, 2017), but require the costly matrix inversion.”
SpinQuant: LLM quantization with learned rotations (2024, unknown): “Detailed GPU latency results are provided in the Appendix.”
A feasible method for optimization with orthogonality constraints (2012, background): “Later, an anonymous referee brought to our attention the work [39] on learning algorithms for neural networks, which also uses the same update.”
A Riemannian geometry with complete geodesics for the set of positive semidefinite matrices of fixed rank (2013, background): “(52) Curves of this type are called quasi-geodesics of order one (Nishimori & Akaho, 2005) or first-order retractions (Absil et al., 2008, Def.”