Computational and statistical tradeoffs via convex relaxation

被引:102
作者
Chandrasekaran, Venkat [1 ,2 ]
Jordan, Michael I. [3 ,4 ]
机构
[1] CALTECH, Dept Comp & Math Sci, Pasadena, CA 91125 USA
[2] CALTECH, Dept Elect Engn, Pasadena, CA 91125 USA
[3] Univ Calif Berkeley, Dept Stat, Berkeley, CA 94720 USA
[4] Univ Calif Berkeley, Dept Elect Engn & Comp Sci, Berkeley, CA 94720 USA
关键词
convex geometry; convex optimization; high-dimensional statistics; LARGE HIDDEN CLIQUE; SAMPLE COMPLEXITY; CUT;
D O I
10.1073/pnas.1302293110
中图分类号
O [数理科学和化学]; P [天文学、地球科学]; Q [生物科学]; N [自然科学总论];
学科分类号
07 ; 0710 ; 09 ;
摘要
Modern massive datasets create a fundamental problem at the intersection of the computational and statistical sciences: how to provide guarantees on the quality of statistical inference given bounds on computational resources, such as time or space. Our approach to this problem is to define a notion of "algorithmic weakening," in which a hierarchy of algorithms is ordered by both computational efficiency and statistical efficiency, allowing the growing strength of the data at scale to be traded off against the need for sophisticated processing. We illustrate this approach in the setting of denoising problems, using convex relaxation as the core inferential tool. Hierarchies of convex relaxations have been widely used in theoretical computer science to yield tractable approximation algorithms to many computationally intractable tasks. In the current paper, we show how to endow such hierarchies with a statistical characterization and thereby obtain concrete tradeoffs relating algorithmic runtime to amount of data.
引用
收藏
页码:E1181 / E1190
页数:10
相关论文
共 56 条
[1]  
Agarwal A, 2012, ARXIV12080129
[2]  
Alon N, 1998, RANDOM STRUCT ALGOR, V13, P457, DOI 10.1002/(SICI)1098-2418(199810/12)13:3/4<457::AID-RSA14>3.0.CO
[3]  
2-W
[4]   Nuclear norm minimization for the planted clique and biclique problems [J].
Ames, Brendan P. W. ;
Vavasis, Stephen A. .
MATHEMATICAL PROGRAMMING, 2011, 129 (01) :69-89
[5]   HIGH-DIMENSIONAL ANALYSIS OF SEMIDEFINITE RELAXATIONS FOR SPARSE PRINCIPAL COMPONENTS [J].
Amini, Arash A. ;
Wainwright, Martin J. .
ANNALS OF STATISTICS, 2009, 37 (5B) :2877-2921
[6]  
[Anonymous], 2008, P 25 INT C MACHINE L, DOI [10.1145/1390156.1390273, DOI 10.1145/1390156.1390273]
[7]  
[Anonymous], 2008, Advances in Neural Information Processing Systems, DOI DOI 10.7751/mitpress/8996.003.0015
[8]  
[Anonymous], 1979, Computers and Intractablity: A Guide to the Theory of NP-Completeness
[9]  
[Anonymous], 1995, GRADUATE TEXTS MATH
[10]  
[Anonymous], 2004, APPROXIMATION ALGORI