The geometry of hypothesis testing over convex cones: Generalized likelihood tests and minimax radii

We consider a compound testing problem within the Gaussian sequence model in which the null and alternative are specified by a pair of closed, convex cones. Such cone testing problem arise in various applications, including detection of treatment effects, trend detection in econometrics, signal detection in radar processing, and shape-constrained inference in non-parametric statistics. We provide a sharp characterization of the GLRT testing radius up to a universal multiplicative constant in terms of the geometric structure of the underlying convex cones. When applied to concrete examples, this result reveals some interesting phenomena that do not arise in the analogous problems of estimation under convex constraints. In particular, in contrast to estimation error, the testing error no longer depends purely on the problem complexity via a volume-based measure (such as metric entropy or Gaussian complexity), other geometric properties of the cones also play an important role. To address the issue of optimality, we prove information-theoretic lower bounds for minimax testing radius again in terms of geometric quantities. Our general theorems are illustrated by examples including the cases of monotone and orthant cones, and involve some results of independent interest.

[1]  Bodhisattva Sen,et al.  Testing against a linear regression model using ideas from shape‐restricted estimation , 2013, 1311.6849.

[2]  Tim Robertson,et al.  On Testing Monotone Tendencies , 1983 .

[3]  D. J. Bartholomew,et al.  A TEST OF HOMOGENEITY FOR ORDERED ALTERNATIVES. II , 1959 .

[4]  G. Pisier.  Probabilistic methods in the geometry of Banach spaces , 1986 .

[5]  C. Pouet.  Nonparametric Goodness-of-Fit Testing Under Gaussian Models , 2004 .

[6]  Jianqing Fan,et al.  Generalized likelihood ratio statistics and Wilks phenomenon , 2001 .

[7]  S. Chatterjee.  A new perspective on least squares under convex constraint , 2014, 1402.0830.

[8]  Cristina Rueda,et al.  Dominance of Likelihood Ratio Tests under Cone Constraints , 1992 .

[9]  N. Meinshausen.  Sign-constrained least squares estimation for high-dimensional regression , 2012, 1202.0889.

[10]  P. Bartlett,et al.  Local Rademacher complexities , 2005, math/0508275.

[11]  A. Shapiro.  Towards a unified theory of inequality constrained testing in multivariate analysis , 1988 .

[12]  Y. Baraud.  Non-asymptotic minimax rates of testing in signal detection , 2002 .

[13]  Erik V. Nordheim,et al.  Hypothesis tests for normal means constrained by linear inequalities , 1986 .

[14]  Tim Robertson,et al.  On Testing Symmetry and Unimodality , 1986 .

[15]  P. Cochat,et al.  Et al , 2008, Archives de pediatrie : organe officiel de la Societe francaise de pediatrie.

[16]  Mary C. Meyer.  A test for linear versus convex regression function using shape‐restricted regression , 2003 .

[17]  Matthias Hein,et al.  Non-negative least squares for high-dimensional linear models: consistency and sparse recovery without regularization , 2012, 1205.0953.

[18]  A. Tsybakov,et al.  Asymptotically exact nonparametric hypothesis testing in sup-norm and at a fixed point , 2000 .

[19]  C. Borell.  The Brunn-Minkowski inequality in Gauss space , 1975 .

[20]  Yu. I. Ingster.  Minimax Testing of Nonparametric Hypotheses on a Distribution Density in the $L_p$ Metrics , 1987 .

[21]  Martin J. Wainwright,et al.  Sharp minimax bounds for testing discrete monotone distributions , 2016, 2016 IEEE International Symposium on Information Theory (ISIT).

[22]  Vladimir Koltchinskii,et al.  Rademacher penalties and structural risk minimization , 2001, IEEE Trans. Inf. Theory.

[23]  V. Spokoiny,et al.  Minimax Nonparametric Hypothesis Testing: The Case of an Inhomogeneous Alternative , 1999 .

[24]  Edward J. Wegman,et al.  Likelihood Ratio Tests for Order Restrictions in Exponential Families , 1978 .

[25]  Donghui Chen,et al.  Nonnegativity constraints in numerical analysis , 2009, The Birth of Numerical Analysis.

[26]  Tim Robertson,et al.  A Likelihood Ratio Test regarding Two Nested but Oblique Order-Restricted Hypotheses , 1984 .

[27]  Cristina Rueda,et al.  Testing non-oblique hypotheses , 1992 .

[28]  Ronitt Rubinfeld,et al.  Sublinear algorithms for testing monotone and unimodal distributions , 2004, STOC '04.

[29]  Tim Robertson,et al.  On Measuring the Conformity of a Parameter Set to a Trend, with Applications , 1982 .

[30]  Tim Robertson,et al.  Testing for and against an Order Restriction on Multinomial Parameters , 1978 .

[31]  Xiaomi Hu,et al.  Likelihood ratio tests for a class of non-oblique hypotheses , 1994 .

[32]  Jianqing Fan,et al.  Nonparametric inference with generalized likelihood ratio tests , 2007 .

[33]  Peter Buhlmann.  Statistical significance in high-dimensional linear models , 2012, 1202.1377.

[34]  L. Goldstein,et al.  Gaussian Phase Transitions and Conic Intrinsic Volumes: Steining the Steiner Formula , 2014, 1411.6265.

[35]  Erich L. Lehmann.  On likelihood ratio tests , 2006 .

[36]  Bonifacio Salvador,et al.  Anomalies of the Likelihood Ratio Test for Testing Restricted Hypotheses , 1991 .

[37]  Stephen E. Fienberg,et al.  Testing Statistical Hypotheses , 2005 .

[38]  O. Besson,et al.  Adaptive Detection of a Signal Whose Signature Belongs to a Cone , 2006, Fourth IEEE Workshop on Sensor Array and Multichannel Processing, 2006..

[39]  M. Ermakov.  Minimax Detection of a Signal In a Gaussian White Noise , 1991 .

[40]  Joel A. Tropp,et al.  Living on the edge: phase transitions in convex programs with random data , 2013, 1303.6672.

[41]  L. Scharf,et al.  Statistical Signal Processing: Detection, Estimation, and Time Series Analysis , 1991 .

[42]  工藤 昭夫,et al.  A Multivariate Analogue of the One-Sided Testについての一注意 (多次元統計解析の数理的研究) , 1979 .

[43]  N. Reid,et al.  Likelihood , 1993 .

[44]  F. T. Wright,et al.  Order restricted statistical inference , 1988 .

[45]  D. Chetverikov.  TESTING REGRESSION MONOTONICITY IN ECONOMETRIC MODELS , 2012, Econometric Theory.

[46]  Fulvio Gini,et al.  Radar Detection and Classification of Jamming Signals Belonging to a Cone Class , 2008, IEEE Transactions on Signal Processing.

[47]  Joel A. Tropp,et al.  From Steiner Formulas for Cones to Concentration of Intrinsic Volumes , 2013, Discret. Comput. Geom..

[48]  J. Cima,et al.  On weak* convergence in ¹ , 1996 .

[49]  Lang Wu,et al.  The Emperor’s new tests , 1999 .

[50]  H. D. Brunk,et al.  Statistical inference under order restrictions : the theory and application of isotonic regression , 1973 .