rbnet.pcfg

Classes

AbstractedPCFG(non_terminals, terminals, ...)

An AbstractedPCFG defines an RBN that has only one non-terminal and one terminal variable, both being discrete with a cardinality corresponding to the number of non-terminal and terminal symbols of the PCFG, respectively.

DiscreteBinaryNonTerminalTransition(weights)

A binary non-terminal transition for discrete non-terminal variables.

DiscreteCell(variable, weights, transitions)

A cell for a discrete variable, where the structural distribution may depend on the variable's value

DiscreteNonTermVar(cardinality[, chart_type])

A discrete non-terminal variable with cardinality cardinality.

DiscretePrior(struc_weights, prior_weights)

A prior distribution over discrete non-terminal variables.

DiscreteTerminalTransition(weights[, ...])

A binary terminal transition for discrete non-terminal and terminal variables.

ExpandedPCFG(non_terminals, terminals, ...)

An ExpandedPCFG...

PCFG(cells, prior, terminal_indices, ...[, ...])

StaticCell(variable, weights, transitions[, ...])

A cell with a static structural distribution (probabilities over the different transitions)