Distributed Parameter Estimation via Pseudo-likelihood

Abstract

Estimating statistical models within sensor networks requires distributed algorithms, in which both data and computation are distributed across the nodes of the network. We propose a general approach for distributed learning based on combining local estimators defined by pseudo-likelihood components, encompassing a number of combination methods, and provide both theoretical and experimental analysis. We show that simple linear combination or max-voting methods, when combined with second-order information, are statistically competitive with more advanced and costly joint optimization. Our algorithms have many attractive properties including low communication and computational cost and "any-time" behavior.

Appears in Proceedings of the 29th International Conference on Machine Learning (ICML 2012)

References in corpus (1)

• Emergence of scaling in random networks

Cited by in corpus (10)

• Distributed Estimation, Information Loss and Exponential Families
• Marginal Pseudo-Likelihood Learning of Markov Network structures
• Marginal Likelihoods for Distributed Parameter Estimation of Gaussian Graphical Models
• Learning Mixed Graphical Models
• Structure Learning of Contextual Markov Networks using Marginal Pseudo-likelihood
• Approximating faces of marginal polytopes in discrete hierarchical models
• Distributed parameter estimation of discrete hierarchical models via marginal likelihoods
• Distributed Parameter Estimation in Probabilistic Graphical Models
• Linear and Parallel Learning of Markov Random Fields
• A local approach to estimation in discrete loglinear models