检索结果-内蒙古大学图书馆

39th Youth Academic Annual Conference of Chinese-Association-of-Automation (YAC)

作者： Wu, Xunhao Fu, Jun Northeastern Univ State Key Lab Synthet Automat Proc Ind Shenyang Peoples R China

ISBN: (纸本)9798350390780;9798350379228

In the context of time-varying directed graphs under uniformly strong connectivity, this paper studies distributed stochastic optimization problems where the local objective functions and inequality constraints are nonlinear, and the local constraint functions are disturbed by uncertain parameters. The goal is to locate a consensus optimal solution to the optimization problem without intervention from a central agent, where each agent obtains the solution through information exchange with neighbouring agents and local computations. In this article, a distributed random projection gradient (DRPG) algorithm is proposed based on the distributed gradient algorithm, random sampling process and Polyak projection operator. Then, this paper theoretically proves that the DRPG algorithm reaches almost sure convergence. Finally, the convergence of the DRPG algorithm is illustrated by a numerical case study.

关键词： distributed stochastic optimization random projection gradient algorithm almost sure convergence

来源：评论

学校读者我要写书评

暂无评论

An enhanced gradient-tracking bound for distributed online stochastic convex optimization

引用

SIGNAL PROCESSING 2024年 217卷

作者： Alghunaim, Sulaiman A. Yuan, Kun Kuwait Univ Dept Elect Engn Kuwait Kuwait Peking Univ Ctr Machine Learning Res Beijing Peoples R China

Gradient-tracking (GT) based decentralized methods have emerged as an effective and viable alternative method to decentralized (stochastic) gradient descent (DSGD) when solving distributed online stochastic optimization problems. Initial studies of GT methods implied that GT methods have worse network dependent rate than DSGD, contradicting experimental results. This dilemma has recently been resolved, and tighter rates for GT methods have been established, which improves upon *** this work, we establish more enhanced rates for GT methods under the online stochastic convex settings. We present an alternative approach for analyzing GT methods for convex problems and over static graphs. When compared to previous analyses, this approach allows us to establish enhanced network dependent rates.

关键词： distributed stochastic optimization Decentralized learning Gradient-tracking Adapt-then-combine

来源：评论

学校读者我要写书评

暂无评论

distributed SUBGRADIENT-FREE stochastic optimization ALGORITHM FOR NONSMOOTH CONVEX FUNCTIONS OVER TIME-VARYING NETWORKS

引用

SIAM JOURNAL ON CONTROL AND optimization 2019年第4期57卷 2821-2842页

作者： Wang, Yinghui Zhao, Wenxiao Hong, Yiguang Zamani, Mohsen Chinese Acad Sci Acad Math & Syst Sci Key Lab Syst & Control Beijing 100190 Peoples R China Univ Chinese Acad Sci Sch Math Sci Beijing 100049 Peoples R China Univ Newcastle Sch Elect Engn & Comp Callaghan NSW 2308 Australia

In this paper we consider a distributed stochastic optimization problem without gradient/subgradient information for local objective functions and subject to local convex constraints. Objective functions may he nonsmooth and observed with stochastic noises, and the network for the distributed design is time-varying. By adding stochastic dithers to local objective functions and constructing randomized differences motivated by the Kiefer-Wolfowitz algorithm, we propose a distributed subgradient-free algorithm for finding the global minimizer with local observations. Moreover, we prove that the consensus of estimates and global minimization can he achieved with probability one over the time-varying network, and we obtain the convergence rate of the mean average of estimates as well. Finally, we give numerical examples to illustrate the performance of the proposed algorithms.

关键词： distributed stochastic optimization gradient-/subgadient-free algorithm nonsmoothness randomized differences

来源：评论

学校读者我要写书评

暂无评论

Diffusion stochastic optimization for Min-Max Problems

引用

IEEE TRANSACTIONS ON SIGNAL PROCESSING 2025年 73卷 259-274页

作者： Cai, Haoyuan Alghunaim, Sulaiman A. Sayed, Ali H. Ecole Polytech Fed Lausanne EPFL Sch Engn CH-1015 Lausanne Switzerland Kuwait Univ Dept Elect Engn Kuwait 12037 Kuwait

The optimistic gradient method is useful in addressing minimax optimization problems. Motivated by the observation that the conventional stochastic version suffers from the need for a large batch size on the order of O(epsilon(-2)) to achieve an epsilon-stationary solution, we introduce and analyze a new formulation termed Diffusion stochastic Same-Sample Optimistic Gradient (DSS-OG). We prove its convergence and resolve the large batch issue by establishing a tighter upper bound, under the more general setting of nonconvex Polyak-Lojasiewicz (PL) risk functions. We also extend the applicability of the proposed method to the distributed scenario, where agents communicate with their neighbors via a left-stochastic protocol. To implement DSS-OG, we can query the stochastic gradient oracles in parallel with some extra memory overhead, resulting in a complexity comparable to its conventional counterpart. To demonstrate the efficacy of the proposed algorithm, we conduct tests by training generative adversarial networks.

关键词： Convergence optimization Training Gradient methods Stability criteria Signal processing algorithms Noise Complexity theory Accuracy Wireless communication Minimax optimization nonconvex Polyak-Lojasiewicz distributed stochastic optimization stochastic optimistic gradient diffusion strategy

来源：评论

学校读者我要写书评

暂无评论

distributed stochastic energy management for ship integrated energy system with uncertain load demand

引用

OCEAN ENGINEERING 2024年第Part3期312卷

作者： Teng, Fei Ban, Zixiao Li, Tieshan Shan, Qihe Li, Yushuai Dalian Maritime Univ Coll Marine Elect Engn Dalian 116026 Peoples R China Univ Elect Sci & Technol China Sch Automat Engn Chengdu 611731 Sichuan Peoples R China Dalian Maritime Univ Nav Coll Dalian 116026 Liaoning Peoples R China Aalborg Univ Dept Comp Sci DK-9220 Aalborg Denmark

Under complex sea conditions, the energy demand for each device of renewable-energy ships presents a random situation, which makes the complex energy demand of the ship integrated energy system (SIES) uncertain during ship navigation. To ensure the economical, stable, and efficient operation of the SIES, this paper proposes a distributed stochastic energy management method to solve the energy management problem (EMP). Firstly, a framework for the SIES including both renewable energy and traditional energy is constructed. Based on the energy efficiency operation index (EEOI) and the operation mode of energy supply devices during navigation, the EMP of the SIES is raised. Then, considering the distributed structure and limited computing resources of the SIES, a distributed stochastic energy management method is proposed. Through this method, the disturbances of load demand can be effectively suppressed, and a stable energy supply is provided for devices such as power propellers. Furthermore, it is analyzed that the proposed method can converge to the O(7) ( 7 ) (7 is the fixed step size of the proposed method) neighborhood of the optimal energy management decision in the mean-square-error sense. Finally, the simulation results verify that the mean-square-error-optimal energy management decision of the SIES can be obtained by the proposed method in different scenarios, and the proposed method can solve the EMP of SIES under complex sea conditions.

关键词： Ship integrated energy system Energy management Uncertain load demand distributed stochastic optimization

来源：评论

学校读者我要写书评

暂无评论

Optimal distributed stochastic mirror descent for strongly convex optimization

引用

AUTOMATICA 2018年 90卷 196-203页

作者： Yuan, Deming Hong, Yiguang Ho, Daniel W. C. Jiang, Guoping Nanjing Univ Sci & Technol Sch Automat Nanjing 210094 Jiangsu Peoples R China Nanjing Univ Posts & Telecommun Coll Automat Nanjing 210023 Jiangsu Peoples R China Chinese Acad Sci Acad Math & Syst Sci Key Lab Syst & Control Beijing 100190 Peoples R China City Univ Hong Kong Dept Math Kowloon Hong Kong Peoples R China

In this paper we consider convergence rate problems for stochastic strongly-convex optimization in the non-Euclidean sense with a constraint set over a time-varying multi-agent network. We propose two efficient non-Euclidean stochastic subgradient descent algorithms based on the Bregman divergence as distance-measuring function rather than the Euclidean distances that were employed by the standard distributed stochastic projected subgradient algorithms. For distributed optimization of non-smooth and strongly convex functions whose only stochastic subgradients are available, the first algorithm recovers the best previous known rate of O(ln(T)/T) (where T is the total number of iterations). The second algorithm is an epoch variant of the first algorithm that attains the optimal convergence rate of O(1/T), matching that of the best previously known centralized stochastic subgradient algorithm. Finally, we report some simulation results to illustrate the proposed algorithms. (C) 2018 Elsevier Ltd. All rights reserved.

关键词： distributed stochastic optimization Strong convexity Non-Euclidean divergence Mirror descent Epoch gradient descent Optimal convergence rate

来源：评论

学校读者我要写书评

暂无评论

Operating peer-to-peer electricity markets under uncertainty via learning-based, distributed optimal control

引用

APPLIED ENERGY 2023年 343卷

作者： Tsaousoglou, Georgios Ellinas, Petros Varvarigos, Emmanouel Tech Univ Denmark Dept Appl Math & Comp Sci Lyngby Denmark Natl Tech Univ Athens Inst Commun & Comp Syst Athens Greece

Towards the global endeavor of clean energy transition, there is a rapid development of distributed energy resources installed in the premises of residential or commercial users, enabling them to act as flexible energy prosumers. Empowering prosumers is envisioned as a catalytic development for modern energy economies, with recent research, as well as innovation and policy actions, pointing to the promising direction of decentralized energy markets, where active energy prosumers exchange energy in a decentralized fashion. Despite the vast amount of recent research on prosumer-centric peer-to-peer (p2p) energy markets, only a small subset of studies accounts for managing the inherent uncertainty of prosumers' flexible *** this paper, we consider the problem of controlling the decisions of energy prosumers' within a p2p exchange network. The multi-bilateral economic dispatch is formulated as an optimal control problem. The proposed solution is based on a direct lookahead policy, effectively addressing the issues of dimensionality and local constraint satisfaction. Experimental simulations demonstrate the method's efficiency and the system's behavior. The proposed formulation and method is shown to effectively address the operation of p2p markets under uncertainty, closely tracking the performance of the (full information) optimal-in-hindsight benchmark.

关键词： distributed energy resources Peer-to-peer electricity markets Optimal control distributed stochastic optimization Learn to optimize ADMM

来源：评论

学校读者我要写书评

暂无评论

Excess-Risk of distributed stochastic Learners

引用

IEEE TRANSACTIONS ON INFORMATION THEORY 2016年第10期62卷 5753-5785页

作者： Towfic, Zaid J. Chen, Jianshu Sayed, Ali H. Univ Calif Los Angeles Dept Elect Engn Los Angeles CA 90095 USA MIT Lincoln Lab Lexington MA 02421 USA Microsoft Res Redmond WA 98052 USA

This paper studies the learning ability of consensus and diffusion distributed learners from continuous streams of data arising from different but related statistical distributions. Four distinctive features for diffusion learners are revealed in relation to other decentralized schemes even under left-stochastic combination policies. First, closed-form expressions for the evolution of their excess-risk are derived for strongly convex risk functions under a diminishing step-size rule. Second, using these results, it is shown that the diffusion strategy improves the asymptotic convergence rate of the excess-risk relative to non-cooperative schemes. Third, it is shown that when the in-network cooperation rules are designed optimally, the performance of the diffusion implementation can outperform that of naive centralized processing. Finally, the arguments further show that diffusion outperforms consensus strategies asymptotically and that the asymptotic excess-risk expression is invariant to the particular network topology. The framework adopted in this paper studies convergence in the stronger mean-square-error sense, rather than in distribution, and develops tools that enable a close examination of the differences between distributed strategies in terms of asymptotic behavior, as well as in terms of convergence rates.

关键词： distributed stochastic optimization diffusion strategies consensus strategies centralized processing excess-risk asymptotic behavior convergence rate combination policy

来源：评论

学校读者我要写书评

暂无评论

On distributed online classification in the midst of concept drifts

引用

NEUROCOMPUTING 2013年 112卷 138-152页

作者： Towfic, Zaid J. Chen, Jianshu Sayed, Ali H. Univ Calif Los Angeles Dept Elect Engn Los Angeles CA 90095 USA

In this work, we analyze the generalization ability of distributed online learning algorithms under stationary and non-stationary environments. We derive bounds for the excess-risk attained by each node in a connected network of learners and study the performance advantage that diffusion has over individual non-cooperative processing. We conduct extensive simulations to illustrate the results. (C) 2013 Elsevier B.V. All rights reserved.

关键词： distributed stochastic optimization Diffusion adaptation Non-stationary data Tracking Classification Risk function Loss function Excess risk

来源：评论

学校读者我要写书评

暂无评论

On the Probability Distribution of distributed optimization Strategies

On the Probability Distribution of Distributed Optimization ...

引用

IEEE Global Conference on Signal and Information Processing (GlobalSIP)

作者： Chen, Jianshu Sayed, Ali H. Univ Calif Los Angeles Dept Elect Engn Los Angeles CA 90034 USA

ISBN: (纸本)9781479902484

We study the steady-state probability distribution of diffusion and consensus strategies that employ constant step-sizes to enable continuous adaptation and learning. We show that, in the small step-size regime, the estimation error at each agent approaches a Gaussian distribution. More importantly, the covariance matrix of this distribution is shown to coincide with the error covariance matrix that would result from a centralized stochastic-gradient strategy. The results hold regardless of the connected topology and help clarify the convergence and learning behavior of distributed strategies in an interesting way.

关键词： Diffusion strategy consensus strategy distributed stochastic optimization central limit theorem steady-state performance

来源：评论

学校读者我要写书评

暂无评论

建议与咨询留下您的常用邮箱和电话号码，以便我们向您反馈解决方案和替代方法

时间限定

文献类型

馆藏选择

核心期刊

语言

文献类型

帮助

文字说明：

检索规则说明：

检索范例：

分类表

所选分类

限定检索结果

文献类型

馆藏范围

日期分布

学科分类号

主题

机构

作者

语言

请选择保存的检索档案：

请选择收藏分类：

通借通还

建议与咨询 留下您的常用邮箱和电话号码，以便我们向您反馈解决方案和替代方法

时间限定

文献类型

馆藏选择

核心期刊

语言

文献类型

帮助

文字说明：

检索规则说明：

检索范例：

分类表

所选分类

限定检索结果

文献类型

馆藏范围

日期分布

学科分类号

主题

机构

作者

语言

请选择保存的检索档案： 新增检索档案 确定 取消

请选择收藏分类： 新增自定义分类 确定 取消

通借通还

建议与咨询留下您的常用邮箱和电话号码，以便我们向您反馈解决方案和替代方法

请选择保存的检索档案：

请选择收藏分类：