Local Minimax Complexity of Stochastic Convex Optimization

被引:0
作者
Zhu, Yuancheng [1 ]
Chatterjee, Sabyasachi [2 ]
Duchi, John [3 ]
Lafferty, John [4 ]
机构
[1] Univ Penn, Wharton Stat Dept, Philadelphia, PA 19104 USA
[2] Univ Chicago, Dept Stat, Chicago, IL 60637 USA
[3] Stanford Univ, Dept Elect Engn, Dept Stat, Stanford, CA 94305 USA
[4] Univ Chicago, Dept Comp Sci, Dept Stat, Chicago, IL 60637 USA
来源
ADVANCES IN NEURAL INFORMATION PROCESSING SYSTEMS 29 (NIPS 2016) | 2016年 / 29卷
关键词
D O I
暂无
中图分类号
TP18 [人工智能理论];
学科分类号
081104 ; 0812 ; 0835 ; 1405 ;
摘要
We extend the traditional worst-case, minimax analysis of stochastic convex optimization by introducing a localized form of minimax complexity for individual functions. Our main result gives function-specific lower and upper bounds on the number of stochastic subgradient evaluations needed to optimize either the function or its "hardest local alternative" to a given numerical precision. The bounds are expressed in terms of a localized and computational analogue of the modulus of continuity that is central to statistical minimax analysis. We show how the computational modulus of continuity can be explicitly calculated in concrete cases, and relates to the curvature of the function at the optimum. We also prove a superefficiency result that demonstrates it is a meaningful benchmark, acting as a computational analogue of the Fisher information in statistical estimation. The nature and practical implications of the results are demonstrated in simulations.
引用
收藏
页数:9
相关论文
共 17 条
  • [1] Brown LD, 1996, ANN STAT, V24, P2524
  • [2] A FRAMEWORK FOR ESTIMATION OF CONVEX FUNCTIONS
    Cai, T. Tony
    Low, Mark G.
    [J]. STATISTICA SINICA, 2015, 25 (02) : 423 - 456
  • [3] Minimax bounds for active learning
    Castro, Rui M.
    Nowak, Robert D.
    [J]. IEEE TRANSACTIONS ON INFORMATION THEORY, 2008, 54 (05) : 2339 - 2353
  • [4] Donoho David, 1987, TECHNICAL REPORT, V137
  • [5] DONOHO DL, 1991, ANN STAT, V19, P633, DOI 10.1214/aos/1176348114
  • [6] STATISTICAL ESTIMATION AND OPTIMAL RECOVERY
    DONOHO, DL
    [J]. ANNALS OF STATISTICS, 1994, 22 (01) : 238 - 270
  • [7] Jean-Baptiste H. U., 1993, Convex Analysis and Minimization Algorithms II, Advanced Theory and Bundle Methods
  • [8] Juditsky A., 2014, Stochastic Systems, V4, P44
  • [9] Karp RM, 2007, PROCEEDINGS OF THE EIGHTEENTH ANNUAL ACM-SIAM SYMPOSIUM ON DISCRETE ALGORITHMS, P881
  • [10] Moulines E., 2011, Advances in Neural Information Processing Systems, V24, P451