Markov chain Monte Carlo over Model Structures
نویسندگان
چکیده
Introduction MCMCMS implements a generic framework for constructing Markov chains. It can be used to perform statistical machine learning in a Bayesian framework. It presents a modular, high level approach to performing MCMC simulations over statistical models that can explain observed data [1]. The two main benefits of MCMCMS are its prior-centric construction of the chain, that does away with the need for an explicit proposal, and the fact that complex, crisp and probabilistic, information can be encoded in the prior by use of a high level language. MCMCMS priors can be encoded as SLPs (Stochastic logic programs, [4]) or as DLPs (Distributional). DLPs extend SLPs by allowing on-the-fly computations of the probabilistic labels.
منابع مشابه
Attribute-Level Heterogeneity
Modeling consumer heterogeneity helps practitioners understand market structures and devise effective marketing strategies. In this research the authors study finite mixture specifications for modeling consumer heterogeneity when each regression coefficient has its own finite mixture, that is, an attribute finite mixture model. An important challenge of such an approach to modeling heterogeneit...
متن کاملModels of Random Sparse Eigenmatrices & Bayesian Analysis of Multivariate Structure
We discuss probabilistic models of random covariance structures defined by distributions over sparse eigenmatrices. The decomposition of orthogonal matrices in terms of Givens rotations defines a natural, interpretable framework for defining distributions on sparsity structure of random eigenmatrices. We explore theoretical aspects and implications for conditional independence structures arisin...
متن کاملBayesian partitioning for classi cation and regressionC
In this paper we propose a new Bayesian approach to data modelling. The Bayesian partition model constructs arbitrarily complex regression and classiication surfaces over the design space by splitting the space into an unknown number of disjoint regions. Within each region the data is assumed to be exchangeable and to come from some simple distribution. Using conjugate priors the marginal likel...
متن کاملSpatial count models on the number of unhealthy days in Tehran
Spatial count data is usually found in most sciences such as environmental science, meteorology, geology and medicine. Spatial generalized linear models based on poisson (poisson-lognormal spatial model) and binomial (binomial-logitnormal spatial model) distributions are often used to analyze discrete count data in which spatial correlation is observed. The likelihood function of these models i...
متن کامل