English

The Proxy Step-size Technique for Regularized Optimization on the Sphere Manifold

Optimization and Control 2022-10-19 v1 Robotics

Abstract

We give an effective solution to the regularized optimization problem g(x)+h(x)g (\boldsymbol{x}) + h (\boldsymbol{x}), where x\boldsymbol{x} is constrained on the unit sphere x2=1\Vert \boldsymbol{x} \Vert_2 = 1. Here g()g (\cdot) is a smooth cost with Lipschitz continuous gradient within the unit ball {x:x21}\{\boldsymbol{x} : \Vert \boldsymbol{x} \Vert_2 \le 1 \} whereas h()h (\cdot) is typically non-smooth but convex and absolutely homogeneous, \textit{e.g.,}~norm regularizers and their combinations. Our solution is based on the Riemannian proximal gradient, using an idea we call \textit{proxy step-size} -- a scalar variable which we prove is monotone with respect to the actual step-size within an interval. The proxy step-size exists ubiquitously for convex and absolutely homogeneous h()h(\cdot), and decides the actual step-size and the tangent update in closed-form, thus the complete proximal gradient iteration. Based on these insights, we design a Riemannian proximal gradient method using the proxy step-size. We prove that our method converges to a critical point, guided by a line-search technique based on the g()g(\cdot) cost only. The proposed method can be implemented in a couple of lines of code. We show its usefulness by applying nuclear norm, 1\ell_1 norm, and nuclear-spectral norm regularization to three classical computer vision problems. The improvements are consistent and backed by numerical experiments.

Keywords

Cite

@article{arxiv.2209.01812,
  title  = {The Proxy Step-size Technique for Regularized Optimization on the Sphere Manifold},
  author = {Fang Bai and Adrien Bartoli},
  journal= {arXiv preprint arXiv:2209.01812},
  year   = {2022}
}

Comments

19 pages

R2 v1 2026-06-28T00:43:31.804Z