Package | Description |
---|---|
org.linqs.psl.application.learning.weight.em | |
org.linqs.psl.application.learning.weight.maxlikelihood |
Modifier and Type | Class and Description |
---|---|
class |
ExpectationMaximization
Abstract superclass for implementations of the expectation-maximization
algorithm for learning with latent variables.
|
class |
HardEM
EM algorithm which fits a point distribution to the single most probable
assignment of truth values to the latent variables during the E-step.
|
class |
PairedDualLearner
Learns the parameters of a HL-MRF with latent variables, using a maximum-likelihood
technique that interleaves updates of the parameters and inference steps for
fast training.
|
Modifier and Type | Class and Description |
---|---|
class |
LazyMaxLikelihoodMPE
Voted perception algorithm that does not require a ground model of pre-specified dimensionality.
|
class |
MaxLikelihoodMPE
Learns weights by optimizing the log likelihood of the data using
the voted perceptron algorithm.
|
class |
MaxPiecewisePseudoLikelihood
Learns weights by optimizing the piecewise-pseudo-log-likelihood of the data using
the voted perceptron algorithm.
|
class |
MaxPseudoLikelihood
Learns weights by optimizing the pseudo-log-likelihood of the data using
the voted perceptron algorithm.
|
Copyright © 2018 University of California, Santa Cruz. All rights reserved.