A unifying representation for a class of dependent random measures

Nicholas Foti, Joseph Futoma, Daniel Rockmore, Sinead Williamson
Proceedings of the Sixteenth International Conference on Artificial Intelligence and Statistics, PMLR 31:20-28, 2013.

Abstract

We present a general construction for dependent random measures based on thinning Poisson processes on an augmented space. The framework is not restricted to dependent versions of a specific nonparametric model, but can be applied to all models that can be represented using completely random measures. Several existing dependent random measures can be seen as specific cases of this framework. Interesting properties of the resulting measures are derived and the efficacy of the framework is demonstrated by constructing a covariate-dependent latent feature model and topic model that obtain superior predictive performance.

Cite this Paper

BibTeX
@InProceedings{pmlr-v31-foti13a, title = {A unifying representation for a class of dependent random measures}, author = {Foti, Nicholas and Futoma, Joseph and Rockmore, Daniel and Williamson, Sinead}, booktitle = {Proceedings of the Sixteenth International Conference on Artificial Intelligence and Statistics}, pages = {20--28}, year = {2013}, editor = {Carvalho, Carlos M. and Ravikumar, Pradeep}, volume = {31}, series = {Proceedings of Machine Learning Research}, address = {Scottsdale, Arizona, USA}, month = {29 Apr--01 May}, publisher = {PMLR}, pdf = {http://proceedings.mlr.press/v31/foti13a.pdf}, url = {https://proceedings.mlr.press/v31/foti13a.html}, abstract = {We present a general construction for dependent random measures based on thinning Poisson processes on an augmented space. The framework is not restricted to dependent versions of a specific nonparametric model, but can be applied to all models that can be represented using completely random measures. Several existing dependent random measures can be seen as specific cases of this framework. Interesting properties of the resulting measures are derived and the efficacy of the framework is demonstrated by constructing a covariate-dependent latent feature model and topic model that obtain superior predictive performance.}, note = {Notable paper award} }
Endnote
%0 Conference Paper %T A unifying representation for a class of dependent random measures %A Nicholas Foti %A Joseph Futoma %A Daniel Rockmore %A Sinead Williamson %B Proceedings of the Sixteenth International Conference on Artificial Intelligence and Statistics %C Proceedings of Machine Learning Research %D 2013 %E Carlos M. Carvalho %E Pradeep Ravikumar %F pmlr-v31-foti13a %I PMLR %P 20--28 %U https://proceedings.mlr.press/v31/foti13a.html %V 31 %X We present a general construction for dependent random measures based on thinning Poisson processes on an augmented space. The framework is not restricted to dependent versions of a specific nonparametric model, but can be applied to all models that can be represented using completely random measures. Several existing dependent random measures can be seen as specific cases of this framework. Interesting properties of the resulting measures are derived and the efficacy of the framework is demonstrated by constructing a covariate-dependent latent feature model and topic model that obtain superior predictive performance. %Z Notable paper award
RIS
TY - CPAPER TI - A unifying representation for a class of dependent random measures AU - Nicholas Foti AU - Joseph Futoma AU - Daniel Rockmore AU - Sinead Williamson BT - Proceedings of the Sixteenth International Conference on Artificial Intelligence and Statistics DA - 2013/04/29 ED - Carlos M. Carvalho ED - Pradeep Ravikumar ID - pmlr-v31-foti13a PB - PMLR DP - Proceedings of Machine Learning Research VL - 31 SP - 20 EP - 28 L1 - http://proceedings.mlr.press/v31/foti13a.pdf UR - https://proceedings.mlr.press/v31/foti13a.html AB - We present a general construction for dependent random measures based on thinning Poisson processes on an augmented space. The framework is not restricted to dependent versions of a specific nonparametric model, but can be applied to all models that can be represented using completely random measures. Several existing dependent random measures can be seen as specific cases of this framework. Interesting properties of the resulting measures are derived and the efficacy of the framework is demonstrated by constructing a covariate-dependent latent feature model and topic model that obtain superior predictive performance. N1 - Notable paper award ER -
APA
Foti, N., Futoma, J., Rockmore, D. & Williamson, S.. (2013). A unifying representation for a class of dependent random measures. Proceedings of the Sixteenth International Conference on Artificial Intelligence and Statistics, in Proceedings of Machine Learning Research 31:20-28 Available from https://proceedings.mlr.press/v31/foti13a.html. Notable paper award

Related Material