Dependency Networks for Inference, Collaborative Filtering, and Data Visualization

By David E. Heckerman, David Maxwell Chickering, Christopher Meek, Robert Rounthwaite, Carl Myers Kadie. 2000

643 citations recorded by the source corpus.

Subfields: Probabilistic & Bayesian Methods

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Probabilistic & Bayesian Methods

Citations in the corpus

Deep Generative Stochastic Networks Trainable by Backprop (2013, background): “6 – Dependency networks: Finally, an unexpected result falls out of the GSN theory: it allows us to provide a novel justification for dependency networks (Heckerman et al., 2000) and for the first time define a proper joint distribution between all the visible variables that is learned by such model”
Generalized Denoising Auto-Encoders as Generative Models (2013, background): “In that respect, the situation is similar to the sampling procedure for dependency networks (Heckerman et al., 2000), in that the pairs (Xt, X̃t−1) are not guaranteed to have the same asymptotic distribution as the pairs (Xt, X̃t) as t → ∞.”
GSNs : Generative Stochastic Networks (2015, methodology): “Dependency networks: Finally, an unexpected result falls out of the GSN theory: it allows us to provide a novel justification for dependency networks (Heckerman et al., 2000) and for the first time define a proper joint distribution between all the visible variables that is learned by such models (S”
Natural Language Processing (Almost) from Scratch (2011, background): “Toutanova et al. (2003), who use maximum entropy classifiers, and a bidirectional dependency network (Heckerman et al., 2001) at inference, reach 97.24% per-word accuracy.”
Deep Learning of Representations: Looking Forward (2013, methodology): “In that respect, the situation is similar to the sampling procedure for dependency networks (Heckerman et al., 2000), in that the pairs (Xt, X̃t−1) are not guaranteed to have the same asymptotic distribution as the pairs (Xt, X̃t) as t → ∞.”
An MDP-Based Recommender System (2002, neutral): “The main model to which we compared our results is the Predictor tool developed by Microsoft as a part of Microsoft Commerce Server 2002, based on the models of (Heckerman et al., 2000).”
Iterative Neural Autoregressive Distribution Estimator NADE-k (2014, background): “Recent work on generative stochastic networks (GSNs), which include denoising auto-encoders as special cases, justifies dependency networks (Heckerman et al., 2000) as well as generalized pseudo-log-likelihood (Goodfellow et al.”
High-dimensional graphs and variable selection with the Lasso (2006, background): “Graph estimates that apply this strategy for complexity reduction are sometimes called dependency networks [9].”
Parameterization and Bayesian Modeling (2004, background): “This “ordered pseudo-Gibbs sampler” (Heckerman et al., 2000) is a Markov chain algorithm, and if the values of the parameters are recorded at the end of each loop of iterations, then they will converge to a distribution.”
Learning relational dependency networks in hybrid domains (2015, methodology): “A dependency network (DN) (Heckerman et al. 2001) is a (cyclic) directed probabilistic model that approximates a joint probability distribution over a set of random variables with a set of conditional probability distributions (CPDs).”
Classification in Networked Data: A Toolkit and a Univariate Case Study (2007, positive): “Relational dependency networks (RDNs, Neville and Jensen, 2003, 2004, forthcoming), extend dependency networks (Heckerman et al., 2000) in much the same way that RBNs extend Bayesian networks.”
A bias/variance decomposition for models using collective inference (2008, background): “RDNs (Neville and Jensen, 2004) extend dependency networks (Heckerman et al., 2000) to work with relational data in much the same way that RMNs extend Markov networks.”
Collaborative Filtering Recommender Systems (2007, background): “Association Rule Mining Association mining techniques build models based on commonly occurring patterns in the ratings matrix [23, 37].”
Suboptimal behavior of Bayes and MDL in classification under misspecification (2004, background): “but one assumes a different noise rate in each leaf of the decision tree [Heckerman et al., 2000]; thus, the transformation is done locally for each leaf rather than globally for the whole hypothesis.”
Collaborative filtering via gaussian probabilistic latent semantic analysis (2003, methodology): “Model-based techniques have been investigated before, most notably Bayesian and non-Bayesian clustering techniques [5, 6, 7, 8], Bayesian networks [5], and dependency networks [9].”