Spatial modeling of brain connectivity data via latent distance models with nodes clustering

Aliverti, Emanuele; Durante, Daniele

doi:10.1002/sam.11412

Cited by 10 publications

(7 citation statements)

References 34 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…a scaled isotropic Gaussian or multivariate uniform). This approach is implemented in the popular R package latentnet (Krivitsky & Handcock, 2008); it is still widely used today (Fosdick et al, 2018;Aliverti & Durante, 2019;.…”

Section: Existing Computational Strategies For Bayesian Inference Of ...mentioning

confidence: 99%

Faster MCMC for Gaussian latent position network models

Spencer

Junker

Sweet

2022

Net Sci

View full text Add to dashboard Cite

Latent position network models are a versatile tool in network science; applications include clustering entities, controlling for causal confounders, and defining priors over unobserved graphs. Estimating each node’s latent position is typically framed as a Bayesian inference problem, with Metropolis within Gibbs being the most popular tool for approximating the posterior distribution. However, it is well-known that Metropolis within Gibbs is inefficient for large networks; the acceptance ratios are expensive to compute, and the resultant posterior draws are highly correlated. In this article, we propose an alternative Markov chain Monte Carlo strategy—defined using a combination of split Hamiltonian Monte Carlo and Firefly Monte Carlo—that leverages the posterior distribution’s functional form for more efficient posterior computation. We demonstrate that these strategies outperform Metropolis within Gibbs and other algorithms on synthetic networks, as well as on real information-sharing networks of teachers and staff in a school district.

show abstract

Section: Existing Computational Strategies For Bayesian Inference Of ...mentioning

confidence: 99%

Faster MCMC for Gaussian latent position network models

Spencer

Junker

Sweet

2022

Net Sci

View full text Add to dashboard Cite

show abstract

“…This approach is still widely-used today (e.g. Fosdick et al [2018], Aliverti and Durante [2019], ); it is also implemented in the popular R package latentnet Krivitsky and Handcock, 2008].…”

Section: Bayesian Inference For Lpmsmentioning

confidence: 99%

Faster MCMC for Gaussian Latent Position Network Models

Spencer¹,

Junker²,

Sweet³

2020

Preprint

View full text Add to dashboard Cite

Latent position network models are a versatile tool in network science; applications include clustering entities, controlling for causal confounders, and defining priors over unobserved graphs. Estimating each node's latent position is typically framed as a Bayesian inference problem, with Metropolis within Gibbs being the most popular tool for approximating the posterior distribution. However, it is well-known that Metropolis within Gibbs is inefficient for large networks; the acceptance ratios are expensive to compute, and the resultant posterior draws are highly correlated. In this article, we propose an alternative Markov chain Monte Carlo strategy-defined using a combination of split Hamiltonian Monte Carlo and Firefly Monte Carlothat leverages the posterior distribution's functional form for more efficient posterior computation. We demonstrate that these strategies outperform Metropolis within Gibbs and other algorithms on synthetic networks, as well as on real informationsharing networks of teachers and staff in a school district.

show abstract

“…Geometry induced by similarity has been often used for defining statistical models of social networks [26,33], as well as networks from other domains, including protein-protein interaction and brain networks [3,25] or the Internet [11]. Moreover, geometric representations of social systems have a long tradition in sociology [8,13,14,39].…”

Section: Introductionmentioning

confidence: 99%

Structural complementarity and similarity: linking relational principles to network motifs

Talaga¹,

Nowak²

2022

Preprint

View full text Add to dashboard Cite

Similarity is one of the most important relational principles shaping the structure of networks. It is linked to homophily, triadic closure and the abundance of triangles as its characterstic motif. However, many important phenomena, such as collaboration and division of labor, are driven rather by complementarity, or differences and synergy. The principle of complementarity attracted much less attention and in particular has not been operationalized in network terms. Here we address this problem and show that complementarity is linked to the abundance of quadrangles and the presence of dense bipartite-like subgraphs. Starting from very general geometric arguments we define two families of coefficients: (1) similarity coefficients generalizing the notions of local clustering and closure; (2) analogous complementarity coefficients based on quadrangles instead of triangles. We study their main properties and behavior in random graph models and demonstrate their theoretical validity and practical utility on a set of real-world social networks. We show that friendship relations are linked to increased levels of structural similarity and health advice to significantly higher complementarity.

show abstract

Spatial modeling of brain connectivity data via latent distance models with nodes clustering

Cited by 10 publications

References 34 publications

Faster MCMC for Gaussian latent position network models

Faster MCMC for Gaussian latent position network models

Faster MCMC for Gaussian Latent Position Network Models

Structural complementarity and similarity: linking relational principles to network motifs

Contact Info

Product

Resources

About