What question did this study set out to answer?

To develop a distribution-free framework for predicting multiple missing links in partially observed networks while rigorously controlling the false discovery rate under unknown missingness patterns.

August 20, 2026Journal of Computational and Graphical StatisticsOpen Access

Conformal network link prediction with false discovery rate control under unstructured missingness

View Full Paper

Ask AI

Bookmark

Authors

WDWenqin Du WMWanteng Ma XDXia Dong

Discussion

Member takes

Overview

Methodological study demonstrates distribution-free false discovery rate control for link prediction in partially observed networks, indicating robust inference under unstructured missingness.

Key Points

To develop a distribution-free framework for predicting multiple missing links in partially observed networks while rigorously controlling the false discovery rate under unknown missingness patterns.
Leveraged the exchangeability structure of weighted graphon models combined with a multi-splitting strategy to construct conformal p-values for row-wise false discovery rate control.
Developed an e-value aggregation scheme to account for arbitrary prediction dependencies and maintain network-wide false discovery rate control.
Evaluated finite-sample theoretical properties and tested the procedure across synthetic simulations and real-world network benchmarks.
Provided finite-sample theoretical guarantees for valid false discovery rate control across weighted, unweighted, undirected, and bipartite networks without requiring missing-rate assumptions.
Confirmed through extensive simulations and empirical network analysis that the framework maintains target error rates while effectively recovering missing connections.

Cite This Study

Du et al. (2026) studied this question.

synapsesocial.com/papers/6a86b4cb8a91293e6a1cc441 https://doi.org/10.1080/10618600.2026.2719794

View Full Paper

Ask AI

Bookmark

August 20, 2026Journal of Computational and Graphical StatisticsOpen Access

Conformal network link prediction with false discovery rate control under unstructured missingness

View Full Paper

Ask AI

Bookmark

Authors

WDWenqin Du WMWanteng Ma XDXia Dong

Discussion

Member takes

Overview

Methodological study demonstrates distribution-free false discovery rate control for link prediction in partially observed networks, indicating robust inference under unstructured missingness.

Key Points

To develop a distribution-free framework for predicting multiple missing links in partially observed networks while rigorously controlling the false discovery rate under unknown missingness patterns.
Leveraged the exchangeability structure of weighted graphon models combined with a multi-splitting strategy to construct conformal p-values for row-wise false discovery rate control.
Developed an e-value aggregation scheme to account for arbitrary prediction dependencies and maintain network-wide false discovery rate control.
Evaluated finite-sample theoretical properties and tested the procedure across synthetic simulations and real-world network benchmarks.
Provided finite-sample theoretical guarantees for valid false discovery rate control across weighted, unweighted, undirected, and bipartite networks without requiring missing-rate assumptions.
Confirmed through extensive simulations and empirical network analysis that the framework maintains target error rates while effectively recovering missing connections.

Cite This Study

Du et al. (2026) studied this question.

synapsesocial.com/papers/6a86b4cb8a91293e6a1cc441 https://doi.org/10.1080/10618600.2026.2719794

View Full Paper

Ask AI

Bookmark