Bilinear Mixed-Effects Models for Dyadic Data

By Peter D. Hoff. 2005

444 citations recorded by the source corpus.

Subfields: Artificial Intelligence, Statistics and Probability

View Paper: https://www.semanticscholar.org/paper/fc129d75d6d33ab425698b5a841273e7f894638b

How researchers see this paper

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

In Artificial Intelligence

In Statistics and Probability

Citations in the corpus

Handbook of Big Data (2016, methodology): “Since latent space models are best fit using Bayesian methodology (Hoff, 2005, 2008), we can easily construct a posterior predictive p-value using the techniques described in Thompson and Geyer (2007). Testing procedures have been recently developed for more complicated models: Yang et al. (2014) pr”
Cluster–Robust Variance Estimation for Dyadic Data (2013, methodology): “ speed dating experiment (Fisman et al., 2006). Current statistical approaches to handling dyadic clustering include the use of parametric restrictions in mixed-effects models (Gelman and Hill, 2007; Hoff, 2005; Kenny, Kashy and Cook, 2006), the spatial error-lag model (Beck, Gleditsch and Beardsley”
GLMix: Generalized Linear Mixed Models For Large-Scale Response Prediction (2016, methodology): “Traditionally the model fitting approaches for GLMix have relied on Gibbs sampling [3, 10] or trying to integrate the log likelihood over the random effect and obtaining the marginal maximum likelihood estimates of the hyper-parameters [4, 8].”
The analysis of social networks (2008, methodology): “…MCMC was not needed to fit the Edges model as logistic regression yields the MLE A related generalized bilinear mixed-effect model developed by Hoff (2005) also assumes conditional independence among ties, but uses an inner-product specification for the effect of the latent quantitative…”
Simulation of the Matrix Bingham–von Mises–Fisher Distribution, With Applications to Multivariate and Relational Data (2007, methodology): “Hoff (2005) discussed parameter estimation for a version of this model, and Hoff and Ward (2004) used such a model to describe networks of international relations.”
Link Prediction via Matrix Factorization (2011, methodology): “Random effects [16] Optional No No”
Model Averaging and Dimension Selection for the Singular Value Decomposition (2006, methodology): “Gabriel (1998) considered maximum-likelihood estimation for a variant of this model in situations where the dimension of D is fixed, and Hoff (2005) considered a symmetric version of this model for the analysis of social network data.”
Representing Degree Distributions, Clustering, and Homophily in Social Networks With Latent Cluster Random Effects Models (2009, unknown): “We have introduced an extension to the latent space model of Hoff et al. (2002) and the latent position clustering model of HRT that also models heterogeneity in actor sociality levels by including random effects.”
Statistical Models for Social Networks (2011, background): “Euclidean Space Hoff et al. (2002) proposed to specify the latent metric as a Euclidean distance – in practice with a dimension K = 2, sometimes K = 3.”
Divide-and-Conquer Matrix Factorization (2011, background): “…applications, including collaborative ﬁltering for recommender systems (e.g., [19] and the references therein), link prediction for social networks [15], click prediction for web search [6], video surveillance [2], graphical model selection [4], document modeling [27], and image alignment [33].”
A state-space mixed membership blockmodel for dynamic network tomography (2008, background): “The latent space models (LSM) instead project nodes onto a latent space, where their similarities can be visualized and explored (Hoff et al., 2002; Hoff, 2003; Handcock et al., 2007).”
Modeling homophily and stochastic equivalence in symmetric relational data (2007, background): “A less general positive semi-definite version of the eigenmodel has been studied by Hoff [2005], in which Λ was taken to be the identity matrix.”
Latent factor models with additive and hierarchically-smoothed user preferences (2013, background): “The final component in computing zui is thus ui which is a random noise that accounts for interactions not explained by the observed attributes (which is termed as a random effect model in the statistics literature [10]).”