Connections Between Score Matching, Contrastive Divergence, and Pseudolikelihood for Continuous-Valued Variables

Many probabilistic models are only defined up to a normalization constant. This makes maximum likelihood estimation of the model parameters very difficult. Typically, one then has to resort to Markov Chain Monte Carlo methods, or approximations of the ...

The random variable simulation problem consists in using a $\boldsymbol {k}$ -dimensional i.i.d. random vector $\boldsymbol {X}^{\boldsymbol {k}}$ with distribution $\boldsymbol {P}_{\boldsymbol {X}}^{\boldsymbol {k}}$ to simulate an $\boldsymbol {n}$ -dimensional i.i.d. random vector $\boldsymbol {Y}^{\boldsymbol {n}}$ so that its distribution is approximately $\boldsymbol {Q}_{\boldsymbol {Y}}^{\boldsymbol {n}}$ . In contrast to previous works, in this paper, ...

Learning undirected graphical models such as Markov random fields is an important machine learning task with applications in many domains. Since it is usually intractable to learn these models exactly, various approximate learning techniques have been ...