ON COMBINATORIAL TESTING PROBLEMS

被引:70
作者
Addario-Berry, Louigi [1 ]
Broutin, Nicolas [2 ]
Devroye, Luc [3 ]
Lugosi, Gabor [4 ,5 ]
机构
[1] McGill Univ, Dept Math & Stat, Montreal, PQ H3A 2K6, Canada
[2] INRIA Rocquencourt, F-78153 Le Chesnay, France
[3] McGill Univ, Sch Comp Sci, Montreal, PQ H3A 2A7, Canada
[4] ICREA, Barcelona, Spain
[5] Pompeu Fabra Univ, Dept Econ, Barcelona, Spain
基金
加拿大自然科学与工程研究理事会;
关键词
Hypothesis testing; multiple hypotheses; Gaussian processes; HIGH DIMENSION; UNIFORM;
D O I
10.1214/10-AOS817
中图分类号
O21 [概率论与数理统计]; C8 [统计学];
学科分类号
020208 ; 070103 ; 0714 ;
摘要
We study a class of hypothesis testing problems in which, upon observing the realization of an n-dimensional Gaussian vector, one has to decide whether the vector was drawn from a standard normal distribution or, alternatively, whether there is a subset of the components belonging to a certain given class of sets whose elements have been "contaminated," that is, have a mean different from zero. We establish some general conditions under which testing is possible and others under which testing is hopeless with a small risk. The combinatorial and geometric structure of the class of sets is shown to play a crucial role. The bounds are illustrated on various examples.
引用
收藏
页码:3063 / 3092
页数:30
相关论文
共 38 条
[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]  
[Anonymous], 1998, Mathematical Methods in Statistics
[5]  
[Anonymous], 2001, Scan Statistics
[6]  
[Anonymous], 1970, CANADIAN MATH MONOGR
[7]  
[Anonymous], 2005, SPRINGER MG MATH
[8]  
Arias-Castro E., 2009, Detection of an abnormal cluster in a network
[9]   Searching for a trail of evidence in a maze [J].
Arias-Castro, Ery ;
Candes, Emmanuel J. ;
Helgason, Hannes ;
Zeitouni, Ofer .
ANNALS OF STATISTICS, 2008, 36 (04) :1726-1757
[10]   SOME NONASYMPTOTIC RESULTS ON RESAMPLING IN HIGH DIMENSION, I: CONFIDENCE REGIONS [J].
Arlot, Sylvain ;
Blanchard, Gilles ;
Roquain, Etienne .
ANNALS OF STATISTICS, 2010, 38 (01) :51-82