TY - GEN

T1 - Inferring with high girth graphical models

AU - Heinemann, Uri

AU - Globerson, Amir

N1 - Publisher Copyright:
Copyright © (2014) by the International Machine Learning Society (IMLS) All rights reserved.

PY - 2014

Y1 - 2014

N2 - Unsupervised learning of graphical models is an important task in many domains. Although maximum likelihood learning is computationally hard, there do exist consistent learning algorithms (e.g., psuedo-likelihood and its variants). However, inference in the learned models is still hard, and thus they are not directly usable. In other words, given a probabilistic query they are not guaranteed to provide an answer that is close to the true one. In the current paper, we provide a learning al-gorithm that is guaranteed to provide approximately correct probabilistic inference. We focus on a particular class of models, namely high girth graphs in the correlation decay regime. It is well known that approximate inference (e.g, using loopy BP) in such models yields marginals that are close to the true ones. Motivated by this, we propose an algorithm that always returns models of this type, and hence in the models it returns inference is approximately correct. We derive finite sample results guaranteeing that beyond a certain sample size, the resulting models will answer probabilistic queries with a high level of accuracy. Results on synthetic data show that the models we learn indeed outperform those obtained by other algorithms, which do not return high girth graphs.

AB - Unsupervised learning of graphical models is an important task in many domains. Although maximum likelihood learning is computationally hard, there do exist consistent learning algorithms (e.g., psuedo-likelihood and its variants). However, inference in the learned models is still hard, and thus they are not directly usable. In other words, given a probabilistic query they are not guaranteed to provide an answer that is close to the true one. In the current paper, we provide a learning al-gorithm that is guaranteed to provide approximately correct probabilistic inference. We focus on a particular class of models, namely high girth graphs in the correlation decay regime. It is well known that approximate inference (e.g, using loopy BP) in such models yields marginals that are close to the true ones. Motivated by this, we propose an algorithm that always returns models of this type, and hence in the models it returns inference is approximately correct. We derive finite sample results guaranteeing that beyond a certain sample size, the resulting models will answer probabilistic queries with a high level of accuracy. Results on synthetic data show that the models we learn indeed outperform those obtained by other algorithms, which do not return high girth graphs.

UR - http://www.scopus.com/inward/record.url?scp=84919820174&partnerID=8YFLogxK

M3 - פרסום בספר כנס

AN - SCOPUS:84919820174

T3 - 31st International Conference on Machine Learning, ICML 2014

SP - 3022

EP - 3041

BT - 31st International Conference on Machine Learning, ICML 2014

PB - International Machine Learning Society (IMLS)

Y2 - 21 June 2014 through 26 June 2014

ER -