Multiple Testing under Dependence via Semiparametric Graphical Models

Jie Liu, Chunming Zhang, Elizabeth Burnside, David Page
; Proceedings of the 31st International Conference on Machine Learning, PMLR 32(2):955-963, 2014.

Abstract

It has been shown that graphical models can be used to leverage the dependence in large-scale multiple testing problems with significantly improved performance (Sun & Cai, 2009; Liu et al., 2012). These graphical models are fully parametric and require that we know the parameterization of f1, the density function of the test statistic under the alternative hypothesis. However in practice, f1 is often heterogeneous, and cannot be estimated with a simple parametric distribution. We propose a novel semiparametric approach for multiple testing under dependence, which estimates f1 adaptively. This semiparametric approach exactly generalizes the local FDR procedure (Efron et al., 2001) and connects with the BH procedure (Benjamini & Hochberg, 1995). A variety of simulations show that our semiparametric approach outperforms classical procedures which assume independence and the parametric approaches which capture dependence.

Cite this Paper


BibTeX
@InProceedings{pmlr-v32-liue14, title = {Multiple Testing under Dependence via Semiparametric Graphical Models}, author = {Jie Liu and Chunming Zhang and Elizabeth Burnside and David Page}, booktitle = {Proceedings of the 31st International Conference on Machine Learning}, pages = {955--963}, year = {2014}, editor = {Eric P. Xing and Tony Jebara}, volume = {32}, number = {2}, series = {Proceedings of Machine Learning Research}, address = {Bejing, China}, month = {22--24 Jun}, publisher = {PMLR}, pdf = {http://proceedings.mlr.press/v32/liue14.pdf}, url = {http://proceedings.mlr.press/v32/liue14.html}, abstract = {It has been shown that graphical models can be used to leverage the dependence in large-scale multiple testing problems with significantly improved performance (Sun & Cai, 2009; Liu et al., 2012). These graphical models are fully parametric and require that we know the parameterization of f1, the density function of the test statistic under the alternative hypothesis. However in practice, f1 is often heterogeneous, and cannot be estimated with a simple parametric distribution. We propose a novel semiparametric approach for multiple testing under dependence, which estimates f1 adaptively. This semiparametric approach exactly generalizes the local FDR procedure (Efron et al., 2001) and connects with the BH procedure (Benjamini & Hochberg, 1995). A variety of simulations show that our semiparametric approach outperforms classical procedures which assume independence and the parametric approaches which capture dependence.} }
Endnote
%0 Conference Paper %T Multiple Testing under Dependence via Semiparametric Graphical Models %A Jie Liu %A Chunming Zhang %A Elizabeth Burnside %A David Page %B Proceedings of the 31st International Conference on Machine Learning %C Proceedings of Machine Learning Research %D 2014 %E Eric P. Xing %E Tony Jebara %F pmlr-v32-liue14 %I PMLR %J Proceedings of Machine Learning Research %P 955--963 %U http://proceedings.mlr.press %V 32 %N 2 %W PMLR %X It has been shown that graphical models can be used to leverage the dependence in large-scale multiple testing problems with significantly improved performance (Sun & Cai, 2009; Liu et al., 2012). These graphical models are fully parametric and require that we know the parameterization of f1, the density function of the test statistic under the alternative hypothesis. However in practice, f1 is often heterogeneous, and cannot be estimated with a simple parametric distribution. We propose a novel semiparametric approach for multiple testing under dependence, which estimates f1 adaptively. This semiparametric approach exactly generalizes the local FDR procedure (Efron et al., 2001) and connects with the BH procedure (Benjamini & Hochberg, 1995). A variety of simulations show that our semiparametric approach outperforms classical procedures which assume independence and the parametric approaches which capture dependence.
RIS
TY - CPAPER TI - Multiple Testing under Dependence via Semiparametric Graphical Models AU - Jie Liu AU - Chunming Zhang AU - Elizabeth Burnside AU - David Page BT - Proceedings of the 31st International Conference on Machine Learning PY - 2014/01/27 DA - 2014/01/27 ED - Eric P. Xing ED - Tony Jebara ID - pmlr-v32-liue14 PB - PMLR SP - 955 DP - PMLR EP - 963 L1 - http://proceedings.mlr.press/v32/liue14.pdf UR - http://proceedings.mlr.press/v32/liue14.html AB - It has been shown that graphical models can be used to leverage the dependence in large-scale multiple testing problems with significantly improved performance (Sun & Cai, 2009; Liu et al., 2012). These graphical models are fully parametric and require that we know the parameterization of f1, the density function of the test statistic under the alternative hypothesis. However in practice, f1 is often heterogeneous, and cannot be estimated with a simple parametric distribution. We propose a novel semiparametric approach for multiple testing under dependence, which estimates f1 adaptively. This semiparametric approach exactly generalizes the local FDR procedure (Efron et al., 2001) and connects with the BH procedure (Benjamini & Hochberg, 1995). A variety of simulations show that our semiparametric approach outperforms classical procedures which assume independence and the parametric approaches which capture dependence. ER -
APA
Liu, J., Zhang, C., Burnside, E. & Page, D.. (2014). Multiple Testing under Dependence via Semiparametric Graphical Models. Proceedings of the 31st International Conference on Machine Learning, in PMLR 32(2):955-963

Related Material