Estimating Latent-Variable Graphical Models using Moments and Likelihoods

Arun Tejasvi Chaganty, Percy Liang
Proceedings of the 31st International Conference on Machine Learning, PMLR 32(2):1872-1880, 2014.

Abstract

Recent work in method of moments provide consistent estimates for latent-variable models, avoiding local optima issues, but these methods can only be applied to certain types of graphical models. In this work, we show that the method of moments in conjunction with a composite marginal likelihood objective yields consistent parameter estimates for a much broader class of directed and undirected graphical models, including loopy graphs with high treewidth. Specifically, we use tensor factorization to reveal partial information about the hidden variables, rendering the otherwise non-convex negative log-likelihood convex. Our approach gracefully extends to models outside our class by incorporating the partial information via posterior regulraization.

Cite this Paper


BibTeX
@InProceedings{pmlr-v32-chaganty14, title = {Estimating Latent-Variable Graphical Models using Moments and Likelihoods}, author = {Chaganty, Arun Tejasvi and Liang, Percy}, booktitle = {Proceedings of the 31st International Conference on Machine Learning}, pages = {1872--1880}, year = {2014}, editor = {Xing, Eric P. and Jebara, Tony}, 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/chaganty14.pdf}, url = {https://proceedings.mlr.press/v32/chaganty14.html}, abstract = {Recent work in method of moments provide consistent estimates for latent-variable models, avoiding local optima issues, but these methods can only be applied to certain types of graphical models. In this work, we show that the method of moments in conjunction with a composite marginal likelihood objective yields consistent parameter estimates for a much broader class of directed and undirected graphical models, including loopy graphs with high treewidth. Specifically, we use tensor factorization to reveal partial information about the hidden variables, rendering the otherwise non-convex negative log-likelihood convex. Our approach gracefully extends to models outside our class by incorporating the partial information via posterior regulraization.} }
Endnote
%0 Conference Paper %T Estimating Latent-Variable Graphical Models using Moments and Likelihoods %A Arun Tejasvi Chaganty %A Percy Liang %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-chaganty14 %I PMLR %P 1872--1880 %U https://proceedings.mlr.press/v32/chaganty14.html %V 32 %N 2 %X Recent work in method of moments provide consistent estimates for latent-variable models, avoiding local optima issues, but these methods can only be applied to certain types of graphical models. In this work, we show that the method of moments in conjunction with a composite marginal likelihood objective yields consistent parameter estimates for a much broader class of directed and undirected graphical models, including loopy graphs with high treewidth. Specifically, we use tensor factorization to reveal partial information about the hidden variables, rendering the otherwise non-convex negative log-likelihood convex. Our approach gracefully extends to models outside our class by incorporating the partial information via posterior regulraization.
RIS
TY - CPAPER TI - Estimating Latent-Variable Graphical Models using Moments and Likelihoods AU - Arun Tejasvi Chaganty AU - Percy Liang BT - Proceedings of the 31st International Conference on Machine Learning DA - 2014/06/18 ED - Eric P. Xing ED - Tony Jebara ID - pmlr-v32-chaganty14 PB - PMLR DP - Proceedings of Machine Learning Research VL - 32 IS - 2 SP - 1872 EP - 1880 L1 - http://proceedings.mlr.press/v32/chaganty14.pdf UR - https://proceedings.mlr.press/v32/chaganty14.html AB - Recent work in method of moments provide consistent estimates for latent-variable models, avoiding local optima issues, but these methods can only be applied to certain types of graphical models. In this work, we show that the method of moments in conjunction with a composite marginal likelihood objective yields consistent parameter estimates for a much broader class of directed and undirected graphical models, including loopy graphs with high treewidth. Specifically, we use tensor factorization to reveal partial information about the hidden variables, rendering the otherwise non-convex negative log-likelihood convex. Our approach gracefully extends to models outside our class by incorporating the partial information via posterior regulraization. ER -
APA
Chaganty, A.T. & Liang, P.. (2014). Estimating Latent-Variable Graphical Models using Moments and Likelihoods. Proceedings of the 31st International Conference on Machine Learning, in Proceedings of Machine Learning Research 32(2):1872-1880 Available from https://proceedings.mlr.press/v32/chaganty14.html.

Related Material