在线双层优化：在线交替梯度法的遗憾分析
Online Bilevel Optimization: Regret Analysis of Online Alternating Gradient Methods.
作者信息
Tarzanagh Davoud Ataee, Nazari Parvin, Hou Bojian, Shen Li, Balzano Laura
机构信息
University of Pennsylvania.
Amirkabir University of Technology.
出版信息
Proc Mach Learn Res. 2024 May;238:2854-2862.
Abstract
This paper introduces an setting in which a sequence of time-varying bilevel problems is revealed one after the other. We extend the known regret bounds for single-level online algorithms to the bilevel setting. Specifically, we provide new notions of , develop an online alternating time-averaged gradient method that is capable of leveraging smoothness, and give regret bounds in terms of the path-length of the inner and outer minimizer sequences.
摘要
本文介绍了一种场景，其中一系列时变双层问题相继出现。我们将单级在线算法的已知遗憾界扩展到双层场景。具体来说，我们提出了新的概念，开发了一种能够利用平滑性的在线交替时间平均梯度方法，并根据内层和外层最小化器序列的路径长度给出遗憾界。
相似文献
1
Online Bilevel Optimization: Regret Analysis of Online Alternating Gradient Methods.在线双层优化：在线交替梯度法的遗憾分析
Proc Mach Learn Res. 2024 May;238:2854-2862.
2
Gradient-Based Framework for Bilevel Optimization of Black-Box Functions: Synergizing Model-Free Reinforcement Learning and Implicit Function Differentiation.基于梯度的黑箱函数双层优化框架：融合无模型强化学习与隐函数微分
Ind Eng Chem Res. 2025 Jan 23;64(5):2831-2844. doi: 10.1021/acs.iecr.4c03584. eCollection 2025 Feb 5.
3
Online Learning Algorithm for Distributed Convex Optimization With Time-Varying Coupled Constraints and Bandit Feedback.具有时变耦合约束和博弈反馈的分布式凸优化在线学习算法
IEEE Trans Cybern. 2022 Feb;52(2):1009-1020. doi: 10.1109/TCYB.2020.2990796. Epub 2022 Feb 16.
4
An Enhanced Memetic Algorithm for Single-Objective Bilevel Optimization Problems.基于改进型 MEMetic 算法的单目标双层优化问题。
Evol Comput. 2017 Winter;25(4):607-642. doi: 10.1162/EVCO_a_00198. Epub 2016 Nov 7.
5
Test problem construction for single-objective bilevel optimization.单目标双层优化的测试问题构建。
Evol Comput. 2014 Fall;22(3):439-77. doi: 10.1162/EVCO_a_00116. Epub 2014 Apr 14.
6
Solving Bilevel Optimization Problems Using Kriging Approximations.使用克里金近似法求解双层优化问题。
IEEE Trans Cybern. 2022 Oct;52(10):10639-10654. doi: 10.1109/TCYB.2021.3061551. Epub 2022 Sep 19.
7
Proximal Online Gradient Is Optimum for Dynamic Regret: A General Lower Bound.
IEEE Trans Neural Netw Learn Syst. 2022 Dec;33(12):7755-7764. doi: 10.1109/TNNLS.2021.3087579. Epub 2022 Nov 30.
8
Beyond Backpropagation: Bilevel Optimization Through Implicit Differentiation and Equilibrium Propagation.
Neural Comput. 2022 Nov 8;34(12):2309-2346. doi: 10.1162/neco_a_01547.
9
An efficient and accurate solution methodology for bilevel multi-objective programming problems using a hybrid evolutionary-local-search algorithm.使用混合进化-局部搜索算法求解双层多目标规划问题的有效且精确的方法。
Evol Comput. 2010 Fall;18(3):403-49. doi: 10.1162/EVCO_a_00015.
10
Stochastic learning via optimizing the variational inequalities.通过优化变分不等式进行随机学习。
IEEE Trans Neural Netw Learn Syst. 2014 Oct;25(10):1769-78. doi: 10.1109/TNNLS.2013.2294741.
引用本文的文献
1
Fair Canonical Correlation Analysis.公平典型相关分析
Adv Neural Inf Process Syst. 2023 Dec;36:3675-3705.
本文引用的文献
1
Investigating Bi-Level Optimization for Learning and Vision From a Unified Perspective: A Survey and Beyond.从统一视角研究学习与视觉的双层优化：综述及展望
IEEE Trans Pattern Anal Mach Intell. 2022 Dec;44(12):10045-10067. doi: 10.1109/TPAMI.2021.3132674. Epub 2022 Nov 7.
2
TADPOLE Challenge: Accurate Alzheimer's disease prediction through crowdsourced forecasting of future data.蝌蚪挑战：通过众包预测未来数据实现阿尔茨海默病的准确预测。
Predict Intell Med. 2019 Oct;11843:1-10. doi: 10.1007/978-3-030-32281-6_1. Epub 2019 Oct 10.
Zotero 插件