A Summary with all its probability on one variable-value Assignment.
A dumb container for messages factor->variable and variable->factor
A collection of marginals inferred by belief propagation.
A Marginal in which all the variables are discrete (either singleton or tensors).
Performs naive mean field inference with a Q Summary that is a set of independent Discrete distributions
A proposal sampler that considers each of the values of a DiscreteVar and scores them efficiently by unrolling factors from the Model just once.
A summary with a separate Proportions distribution for each of its DiscreteVars
A Marginal associated with a Factor.
Manage and use a queue to more often revisit low-scoring factors and re-sample their variables.
Sample a value for a single variable.
A Summary that can be used to gather weighted samples into its Marginals.
Besag's Iterated Conditional Modes.
A Summary with all its probability on one variable-value Assignment, and which can also be used for learning because it includes factors.
A Metropolis-Hastings sampler.
User: apassos Date: 5/29/13 Time: 8:46 AM
Stores a marginal distribution containing a joint distribution over a set of variables.
An inference engine that finds score-maximizing values.
An inferencer for mean field inference.
For storing one of the MCMC sampling proposals considered.
Samplers that generate a list of Proposal objects, and select one log-proportionally to their modelScore.
Samplers that key off of particular contexts.
Instead of randomly sampling according to the distribution, always pick the setting with the maximum acceptanceScore.
Tries each one of the settings in the Iterator provided by the abstract method "settings(C)", scores each, builds a distribution from the scores, and samples from it.
A Summary containing only one Marginal.
The result of inference: a collection of Marginal objects.
A Summary that contains multiple Marginals of type M, each a marginal distribution over a single variable.
Tries each one of the settings of the given variable, scores each, builds a distribution from the scores, and samples from it.
User: apassos Date: 6/15/13 Time: 7:38 AM