近最优的懒惰二阶预言机凸优化
Near-Optimal Convex Optimization with Lazy Second-Order Oracles
浏览论文内容
中文总结 AI 辅助
本文研究懒惰二阶预言机下凸优化的迭代复杂度,证明下界Ω(m+m^{1/7}ε^{-2/7})并提出新方法达到相同上界,显著改进先前结果并紧至对数因子。
中文摘要 AI 辅助
本文研究了使用懒惰二阶预言机(Doikov, Chayti, 和 Jaggi, ICML 2023)进行凸优化的复杂性,其中算法在每次迭代时查询梯度,并每 $m$ 次迭代查询一次 Hessian。在此设置下,我们通过一种新颖的块零链构造,证明了找到 $\u03b5$-解所需的总迭代次数下界为 $\u03a9(m+ m^{1/7} \u03b5^{-2/7})$。随后,我们提出了一种新方法,实现了新的上界 $\tilde{\mathcal{O}}(m+ m^{1/7} \u03b5^{-2/7})$,显著改进了先前(Chen, Liu, Luo, 和 Zhang, COLT 2026)的 $\tilde{\mathcal{O}}(m+ m^{13/21} \u03b5^{-2/7})$,并且在对数因子范围内是紧的。
英文摘要
This paper studies the complexity of convex optimization using lazy second-order oracles (Doikov, Chayti, and Jaggi, ICML 2023), where an algorithm queries gradients every iteration and Hessians once per $m$ iterations. Under this setting, we show a lower bound of $Ω(m+ m^{1/7} ε^{-2/7})$ on the number of total iterations to find an $ε$-solution using a novel block zero-chain construction. Then we propose a novel method that achieves a new upper bound of $\tilde{\mathcal{O}}(m+ m^{1/7} ε^{-2/7})$, which significantly improves the prior one (Chen, Liu, Luo, and Zhang, COLT 2026) of $\tilde{\mathcal{O}}(m+ m^{13/21} ε^{-2/7})$ and is tight up to logarithmic factors.
发表机构
- Tsinghua University(清华大学)
- Westlake University(西湖大学)
机构由 AI 辅助整理,请以论文原文为准。