• Corpus ID: 218517119

Bias-Variance Tradeoffs in Joint Spectral Embeddings

@article{Draves2020BiasVarianceTI,
  title={Bias-Variance Tradeoffs in Joint Spectral Embeddings},
  author={Benjamin Draves and Daniel Lewis Sussman},
  journal={arXiv: Statistics Theory},
  year={2020}
}
Latent position models and their corresponding estimation procedures offer a statistically principled paradigm for multiple network inference by translating multiple network analysis problems to familiar task in multivariate statistics. Latent position estimation is a fundamental task in this framework yet most work focus only on unbiased estimation procedures. We consider the ramifications of utilizing biased latent position estimates in subsequent statistical analysis in exchange for sizable… 

Figures and Tables from this paper

The Importance of Being Correlated: Implications of Dependence in Joint Spectral Inference across Multiple Networks
TLDR
This paper shows that the embedding procedure is flexible and robust, and proves a central limit theorem for this embedding and explicitly compute the limiting covariance, and constructs an appropriately calibrated omnibus embedding that can detect changes in real biological networks that previous embedding procedures could not discern.
Multiple Network Embedding for Anomaly Detection in Time Series of Graphs.
TLDR
Applied to a large-scale commercial search engine time series of graphs, the adaptation of statistically principled methods for joint graph inference, specifically multiple adjacency spectral embedding and omnibus embedding are demonstrated to be applicability and identify the anomalous vertices beyond just large degree change.
Discussion of “Co-citation and Co-authorship Networks of Statisticians” by Pengsheng Ji, Jiashun Jin, Zheng Tracy Ke, and Wanshan Li
TLDR
This discussion focuses on the paper’s particular methods and applications in dynamic network analysis, which will spur further work on multilayer, dynamic and other complex network settings.
Clustered Graph Matching for Label Recovery and Graph Classification
TLDR
It is demonstrated both in theory and practice that if the graphs come from different network classes, then clustering the networks into classes followed by matching the new graph to cluster-averages can yield higher reputation matching performance than matching to the global average graph.
Nonparametric Two-Sample Hypothesis Testing for Random Graphs with Negative and Repeated Eigenvalues.
TLDR
A nonparametric two-sample test statistic is proposed for low-rank, conditionally independent edge random graphs whose edge probability matrices have negative eigenvalues and arbitrarily close eigen Values and it is shown that the test statistic, appropriately scaled, is consistent for sufficiently dense graphs.

References

SHOWING 1-10 OF 69 REFERENCES
The Importance of Being Correlated: Implications of Dependence in Joint Spectral Inference across Multiple Networks
TLDR
This paper shows that the embedding procedure is flexible and robust, and proves a central limit theorem for this embedding and explicitly compute the limiting covariance, and constructs an appropriately calibrated omnibus embedding that can detect changes in real biological networks that previous embedding procedures could not discern.
A central limit theorem for an omnibus embedding of multiple random graphs and implications for multiscale network inference
TLDR
An "omnibus" embedding in which multiple graphs on the same vertex set are jointly embedded into a single space with a distinct representation for each graph is described, which achieves near-optimal inference accuracy and allows the identification of specific brain regions associated with population-level differences.
Inference for Multiple Heterogeneous Networks with a Common Invariant Subspace
TLDR
A new model is introduced, the common subspace independent-edge multiple random graph model, which describes a heterogeneous collection of networks with a shared latent structure on the vertices but potentially different connectivity patterns for each graph, and is both flexible enough to meaningfully account for important graph differences and tractable enough to allow for accurate inference in multiple networks.
A random effects stochastic block model for joint community detection in multiple networks with applications to neuroimaging
TLDR
A modeling framework for joint community detection in a group of related networks, which can be considered as a sample from a population of networks, and an overall putative community structure representative of the group as well as subject-specific variations within each group are developed.
A central limit theorem for an omnibus embedding of random dot product graphs
TLDR
An omnibus embedding in which multiple graphs on the same vertex set are jointly embedded into a single space with a distinct representation for each graph is described, achieving near-optimal inference accuracy when graphs arise from a common distribution and yet retains discriminatory power as a test procedure for the comparison of different graphs.
Connectome Smoothing via Low-Rank Approximations
TLDR
It is demonstrated that the low-rank methods outperform the standard sample mean for a variety of independent edge distributions as well as human connectome data derived from the magnetic resonance imaging, especially when the sample sizes are small.
Law of Large Graphs
TLDR
Theoretical results for the stochastic blockmodel show that this method will offer major improvements when there are many vertices, and results indicate that low-rank methods should be a key part of the tool box for researchers studying populations of graphs.
Consistency of Maximum Likelihood for Continuous-Space Network Models
TLDR
This work establishes three levels of consistency for non-parametric maximum likelihood inference as the number of nodes grows: the estimated locations of all nodes converge in probability on their true locations, the distribution over locations in the latent space converges on the true distribution, and the distributionover graphs of arbitrary size converges.
Hypothesis Testing For Network Data in Functional Neuroimaging
TLDR
This work draws on concepts and techniques from geometry, and high-dimensional statistical inference, for testing in the context of networks derived from functional neuroimaging data on human subjects from the 1000 Functional Connectomes Project, and shows that this global test is more statistical powerful, than a mass-univariate approach.
Tunable Eigenvector-Based Centralities for Multiplex and Temporal Networks
TLDR
A linear-algebraic framework that generalizes eigenvector-based centralities---including PageRank---to provide a common framework for two popular classes of multilayer networks: multiplex networks with layers encoding different types of relationships and temporal networks in which the relationships change in time.
...
...