Composite Likelihood Methods

By Harry Joe, Nancy Reid, Peter Xuekun, David Firth. 2012

730 citations recorded by the source corpus.

View Paper: https://www.semanticscholar.org/paper/bc79768d62cd9cf0e4ffe924dbed4da7ad1db482

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

Citations in the corpus

Genomic Similarity and Kernel Methods II: Methods for Genomic Information (2010, unknown): “…to specify a joint probability density for all attributes, it might be possible to specify a composite pseudo-density that captures some of the main features (e.g. marginal and pair-wise probabilities), similar to how composite likelihoods are used to obtain unbiased parameter estimates [10] .”
A robust pairwise likelihood method for incomplete longitudinal binary data arising in clusters (2011, methodology): “We address this challenge here for the analysis of incomplete clustered longitudinal binary data by adopting the framework of pairwise likelihood (PWL) inference (Lindsay, 1988; Cox & Reid, 2004).”
Consistency by Agreement in Zero-Shot Neural Machine Translation (2019, methodology): “In the statistics literature, this estimation approach is called maximum composite likelihood (Besag, 1975; Lindsay, 1988) as it composes the objective out of (sometimes weighted) terms that represent conditional sub-likelihoods (in our example, Pθ (xj | xi)).”
Statistical inference for max‐stable processes in space and time (2012, methodology): “Composite likelihood estimation for the space–time setting Composite likelihood methods go back to Besag (1974) and Lindsay (1988) and a vast literature is now available, from a theoretical and an applied point of view on the topic. For more information we refer to Varin (2007), who presented an ove”
On the likelihood function of Gaussian max-stable processes (2011, background): “…z(1), . . . , z(N ) from the model f (z; θ) on R K with K fixed, under the usual regularity conditions, θ̂ is a consistent parameter estimator (Lindsay, 1988; Varin & Vidoni, 2005): N 1/2(θ̂ − θ)→ Nq{0, H(θ)−1 J (θ)H(θ)−1} (8) in distribution as N→+∞, where H(θ)= E{−∇2 c(θ)} and J (θ)=…”
A New Estimation Approach to Integrate Latent Psychological Constructs in Choice Modeling (2014, background): “The CML approach, which belongs to the more general class of composite likelihood function approaches (see Lindsay, 1988), may be explained in a simple manner as follows.”
Space–time modelling of extreme events (2012, methodology): “Several models for max-stable processes have been suggested; for example, Smith (1990) proposed a max-stable model with deterministic storm shapes, and Schlather (2002) proposed one based on a truncated Gaussian process.”
Non-Stationary Dependence Structures for Spatial Extremes (2014, methodology): “This explains why pairwise likelihoods (Lindsay, 1988; Varin et al., 2011) have become the standard tool for inference in this context (Padoan et al., 2010; Thibaud et al., 2013; Huser and Davison, 2014), although more efficient approaches based on the point process characterization of extremes…”
Learning Graphical Model Parameters with Approximate Marginal Inference (2013, background): “If defined on exact marginals, this is a type of composite likelihood [40].”
From learning models of natural image patches to whole image restoration (2011, methodology): “A common approximation to learning MRFs is to approximate the log probability of an image as a sum of local marginal or conditional probabilities as in the method of composite likelihood [13] or directed models of images [14].”
Truncated regular vines in high dimensions with application to financial data (2012, background): “1) may be interpreted as a composite likelihood (see Lindsay (1988) and Varin et al.”