Improved Gaussian Mixture Density Estimates Using Bayesian Penalty Terms and Network Averaging

By Dirk Ormoneit, Volker Tresp. 1995

How researchers see this paper

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

Citations in the corpus

A Deep and Tractable Density Estimator (2013, background): “A reliable approach to improve on some given estimator is to instead construct an ensemble of multiple, strong but different estimators, e.g. with bagging (Ormoneit & Tresp, 1996) or stacking (Smyth & Wolpert, 1999).”
Mixtures of Probabilistic Principal Component Analyzers (1999, background): “(Another would be to introduce priors over the parameters of the full covariance matrix, as implemented by Ormoneit and Tresp (1996).) Two common constraints are either to restrict the covariance to be isotropic or to be diagonal.”
SMEM Algorithm for Mixture Models (1998, unknown): “Looking at equation 5 carefully, one can see that the Q function can be represented in the form of a direct sum: Q(Θ|Θ) = M ∑ m=1 qm(θm|Θ ), (7) where qm(θm|Θ ) = N ∑ n=1 P (m|xn; Θ ) log αmpm(xn; θm) (8) and depends only on αm and θm.”
Extreme deconvolution: Inferring complete distribution functions from noisy, heterogeneous and incomplete observations (2009, methodology): “d merge procedure described below. 4.1. Bayesian-inspired regularization. The problem of singular covariances can be mitigated through the use of priors on the model parameters in a Bayesian setting [Ormoneit and Tresp (1996)]. It should be emphasized here that this calculation is only Bayesian in t”
Connectionist multivariate density-estimation and its application to speech synthesis (2016, unknown): “with bagging (Ormoneit and Tresp, 1995) or stacking (Smyth and Wolpert, 1999).”