Bootstrapping Networks with Latent Space Structure

Abstract

A core problem in statistical network analysis is to develop network analogues of classical techniques. The problem of bootstrapping network data stands out as especially challenging, since typically one observes only a single network, rather than a sample. Here we propose two methods for obtaining bootstrap samples for networks drawn from latent space models. The first method generates bootstrap replicates of network statistics that can be represented as U-statistics in the latent positions, and avoids actually constructing new bootstrapped networks. The second method generates bootstrap replicates of whole networks, and thus can be used for bootstrapping any network function. Commonly studied network quantities that can be represented as U-statistics include many popular summaries, such as average degree and subgraph counts, but other equally popular summaries, such as the clustering coefficient, are not expressible as U-statistics and thus require the second bootstrap method. Under the assumption of a random dot product graph, a type of latent space network model, we show consistency of the proposed bootstrap methods. We give motivating examples throughout and demonstrate the effectiveness of our methods on synthetic data.

References in corpus (10)

• Convergence and Concentration of Empirical Measures under Wasserstein Distance in Unbounded Functional Spaces
• A central limit theorem for an omnibus embedding of multiple random graphs and implications for multiscale network inference
• Optimal Change Point Detection and Localization in Sparse Dynamic Networks
• Estimation of subgraph density in noisy networks
• Consistency of Maximum Likelihood for Continuous-Space Network Models I
• Bootstrapping Exchangeable Random Graphs
• Network Representation Using Graph Root Distributions
• A Family of Tractable Graph Distances
• Graph Distance from the Topological View of Non-backtracking Cycles
• Network Dependence Testing via Diffusion Maps and Distance-Based Correlations

Cited by in corpus (9)

• Bootstrapping Exchangeable Random Graphs
• Consistency of Maximum Likelihood for Continuous-Space Network Models I
• Pitfalls of Climate Network Construction: A Statistical Perspective
• Nonparametric two-sample hypothesis testing for low-rank random graphs of differing sizes
• On Two Distinct Sources of Nonidentifiability in Latent Position Random Graph Models
• On the Theoretical Properties of the Network Jackknife
• Estimating Graph Dimension with Cross-validated Eigenvalues
• Valid Two-Sample Graph Testing via Optimal Transport Procrustes and Multiscale Graph Correlation with Applications in Connectomics
• Trading off Accuracy for Speedup: Multiplier Bootstraps for Subgraph Counts