Variational EM Algorithms for Non-Gaussian Latent Variable Models

By Jason A. Palmer, David Paul Wipf, Kenneth Kreutz-Delgado, Bhaskar D. Rao. 2005

209 citations recorded by the source corpus.

Subfields: Generative Models, 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 Generative Models

In Probabilistic & Bayesian Methods

Citations in the corpus

Towards Interpretable Natural Language Understanding with Explanations as Latent Variables (2020, methodology): “Variational EM [8] is used for the optimization, and only a set of human explanations are required for guiding the explanation generation process.”
Composing graphical models with neural networks for structured representations and fast inference (2016, methodology): “First, the fact that there are only a fixed number of hidden units makes it unlikely that the model can [1,2,3] Pr bab listic graphical m els? [4,5,6] [4] Fox, Sudderth, Jord n, illsky.”
Affine Independent Variational Inference (2012, background): “4) is the ‘local’ variational bounding procedure [14, 11, 22, 18].”
FABIA: factor analysis for bicluster acquisition (2010, methodology): “Benchmark datasets published in Prelic et al. (2006) and Li et al.”
A unified Bayesian framework for MEG/EEG source imaging (2009, unknown): “, 2005), automatic relevance determination (ARD) (MacKay, 1992; Neal, 1996; Ramírez, 2005; Ramírez and Makeig, 2006; Tipping, 2001), and several related variational Bayesian methods (Friston et al., 2006, 2008; Nummenmaa et al., 2007; Palmer et al., 2006; Sahani and Nagarajan, 2004; Sato et al., 200”
An Empirical Bayesian Strategy for Solving the Simultaneous Sparse Approximation Problem (2007, background): “While these two strategies can be related in certain settings [35], this topic is beyond the scope of the current work.”
Gaussian Kullback-Leibler approximate inference (2013, methodology): “As detailed in Section 5.1.1, local variational bounding procedures are applicable provided tight exponentiated quadratic lower-bounds to the site projection potentials {φn}Nn=1 exist— that is each site potential is required to be super-Gaussian (Palmer et al., 2006).”
Bayesian reasoning and machine learning (2012, unknown): “1 Local approximation In a local method [219, 149, 119, 228], one replaces f by a suitable function for which the integral can be computed.”
Rectified Factor Networks (2015, methodology): “The performance of RFNs as unsupervised methods for data representation was compared to: (1) RFN: rectified factor networks, (2) RFNn: RFNs without normalization, (3) DAE: denoising autoencoders with rectified linear units, (4) RBM: restricted Boltzmann machines with Gaussian visible units and hidde”
Iterative Reweighted 1 and 2 Methods for Finding Sparse Solutions (2010, methodology): “Using results from [17], it is relatively straightforward to show that log (|x i | + ) ≤ x 2 i γ i + log 2 + 2γ i 1 2 + 2 − 2 + 2γ i 1 2 − 2 4γ i (14) for all , γ i ≥ 0, with equality iff γ i = x 2 i + |x i |.”