paper-with-me

홈 › Papers

Regularized Newton Method with Global $O(1/k^2)$ Convergence

2021-12-03 · Konstantin Mishchenko

We present a Newton-type method that converges fast from any initialization and for arbitrary convex objectives with Lipschitz Hessians. We achieve this by merging the ideas of cubic regularization with a certain adaptive Levenberg--Marquardt penalty. In particular, we show that the iterates given by $x^{k+1}=x^k - \bigl(\nabla^2 f(x^k) + \sqrt{H\|\nabla f(x^k)\|} \mathbf{I}\bigr)^{-1}\nabla f(x^k)$, where $H>0$ is a constant, converge globally with a $\mathcal{O}(\frac{1}{k^2})$ rate. Our method is the first variant of Newton's method that has both cheap iterations and provably fast global convergence. Moreover, we prove that locally our method converges superlinearly when the objective is strongly convex. To boost the method's performance, we present a line search procedure that does not need prior knowledge of $H$ and is provably efficient.

📄 PDF Abstract BibTeX arXiv:2112.02089

Code (2)

konstmish/global-newton 공식 구현
OPTAMI/OPTAMI pytorch

Similar Papers 제목 키워드 기반

Bridging the Gap between Newton-Raphson Method and Regularized Policy Iteration

2023-10-11 · Zeyang Li, Chuxiong Hu, Yunan Wang, Guojian Zhan 외

Regularization is one of the most important techniques in reinforcement learning algorithms. The well-known soft actor-critic algorithm is a special case of regularized policy iteration where the regularizer is chosen as…

Sketch-and-Project Meets Newton Method: Global $\mathcal O(k^{-2})$ Convergence with Low-Rank Updates

2023-05-22 · Slavomír Hanzely

In this paper, we propose the first sketch-and-project Newton method with fast $\mathcal O(k^{-2})$ global convergence rate for self-concordant functions. Our method, SGN, can be viewed in three ways: i) as a sketch-and-…

Krylov Cubic Regularized Newton: A Subspace Second-Order Method with Dimension-Free Convergence Rate

2024-01-05 · Ruichen Jiang, Parameswaran Raman, Shoham Sabach, Aryan Mokhtari 외

Second-order optimization methods, such as cubic regularized Newton methods, are known for their rapid convergence rates; nevertheless, they become impractical in high-dimensional problems due to their substantial memory…

Second-order methodssubspace methods

Gradient Regularized Newton Boosting Trees with Global Convergence

2026-05-01 · Nikita Zozoulenko, Daniel Falkowski, Thomas Cass, Lukas Gonon arxiv

Gradient Boosting Decision Trees (GBDTs) dominate tabular machine learning, with modern implementations like XGBoost, LightGBM, and CatBoost being based on Newton boosting: a second-order descent step in the space of dec…

Simple Stepsize for Quasi-Newton Methods with Global Convergence Guarantees

2025-08-27 · Artem Agafonov, Vladislav Ryspayev, Samuel Horváth, Alexander Gasnikov 외 arxiv

Quasi-Newton methods are widely used for solving convex optimization problems due to their ease of implementation, practical efficiency, and strong local convergence guarantees. However, their global convergence is typic…