On Primal and Dual Approaches for Distributed Stochastic Convex Optimization over Networks

被引:0
作者
Dvinskikh, Darina [1 ,2 ]
Gorbunov, Eduard [3 ]
Gasnikov, Alexander [2 ,3 ,4 ]
Dvurechensky, Pavel [1 ,2 ]
Uribe, Cesar A. [5 ]
机构
[1] Weierstrass Inst Appl Anal & Stochast, Berlin, Germany
[2] Inst Informat Transmiss Problems, Moscow, Russia
[3] Moscow Inst Phys & Technol, Moscow, Russia
[4] Natl Res Univ Higher Sch Econ, Moscow, Russia
[5] MIT, Lab Informat & Decis Syst, Cambridge, MA 02139 USA
来源
2019 IEEE 58TH CONFERENCE ON DECISION AND CONTROL (CDC) | 2019年
基金
俄罗斯科学基金会;
关键词
ASYMPTOTIC AGREEMENT; GRADIENT-METHOD; ALGORITHMS; CONVERGENCE;
D O I
暂无
中图分类号
TP [自动化技术、计算机技术];
学科分类号
0812 ;
摘要
We introduce primal and dual stochastic gradient oracle methods for distributed convex optimization problems over networks. We show that the proposed methods are optimal (in terms of communication steps) for primal and dual oracles. Additionally, for a dual stochastic oracle, we propose a new analysis method for the rate of convergence in terms of duality gap and probability of large deviations. This analysis is based on a new technique that allows to bound the distance between the iteration sequence and the optimal point. By the proper choice of batch size, we can guarantee that this distance equals (up to a constant) to the distance between the starting point and the solution.
引用
收藏
页码:7435 / 7440
页数:6
相关论文
共 44 条
[1]  
Abadi M., 2016, C LANG RES EV MARC, P3243
[2]   Local-area mobility support through cooperating hierarchies of Mobile IP foreign agents [J].
Abdel-Hamid, A ;
Abdel-Wahab, H .
PROCEEDINGS OF THE SIXTH IEEE SYMPOSIUM ON COMPUTERS AND COMMUNICATIONS, 2001, :479-484
[3]   Dual approaches to the minimization of strongly convex functionals with a simple structure under affine constraints [J].
Anikin, A. S. ;
Gasnikov, A. V. ;
Dvurechensky, P. E. ;
Tyurin, A. I. ;
Chernov, A. V. .
COMPUTATIONAL MATHEMATICS AND MATHEMATICAL PHYSICS, 2017, 57 (08) :1262-1276
[4]  
[Anonymous], FOUND TRENDS MACH LE
[5]  
[Anonymous], ARXIV180506045
[6]  
Bogolubsky L, 2016, Adv. Neural Inf. Process. Syst., P4914
[7]   ASYMPTOTIC AGREEMENT IN DISTRIBUTED ESTIMATION [J].
BORKAR, V ;
VARAIYA, PP .
IEEE TRANSACTIONS ON AUTOMATIC CONTROL, 1982, 27 (03) :650-655
[8]   Large-Scale Machine Learning with Stochastic Gradient Descent [J].
Bottou, Leon .
COMPSTAT'2010: 19TH INTERNATIONAL CONFERENCE ON COMPUTATIONAL STATISTICS, 2010, :177-186
[9]   TOSS: Tailoring Online Server Systems through Binary Feature Customization [J].
Chen, Yurong ;
Sun, Shaowen ;
Lan, Tian ;
Venkataramani, Guru .
FEAST'18: PROCEEDINGS OF THE 2018 WORKSHOP ON FORMING AN ECOSYSTEM AROUND SOFTWARE TRANSFORMATION, 2018, :1-7
[10]   Fast Primal-Dual Gradient Method for Strongly Convex Minimization Problems with Linear Constraints [J].
Chernov, Alexey ;
Dvurechensky, Pavel ;
Gasnikov, Alexander .
DISCRETE OPTIMIZATION AND OPERATIONS RESEARCH, DOOR 2016, 2016, 9869 :391-403