摘要：The general functional form of composite likelihoods is derived by minimizing the Kullback-Leibler distance under structural constraints associated with low dimensional densities. Connections with the I-projection and the maximum entropy distributions are shown. Asymptotic properties of composite likelihood inference under the proposed information-theoretical framework are established.
关键词：Composite Likelihood; I-Divergence; Information Theory; Likelihood Weights; Maximum Entropy Distribution
