Conformal Graph Prediction with Z-Gromov Wasserstein Distances

Gabriel Melo, Thibaut de Saivre, Anna Calissano, Florence d’Alché-Buc · 2026

Read Conformal Graph Prediction with Z-Gromov Wasserstein Distances by Gabriel Melo, Thibaut de Saivre, Anna Calissano, Florence d’Alché-Buc (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2603-02460.txt.

Description

Gabriel Melo, Thibaut de Saivre, Anna Calissano, and Florence d'Alché-Buc add error bounds to graph prediction in this 2026 paper. Some tasks output whole graphs, not single numbers. Honest error ranges for such outputs have been scarce. Their method gives coverage promises that need no assumptions about the data spread.

A distance between graphs anchors the method's mismatch score. That distance compares graphs no matter how their nodes are ordered. A quantile method then adapts the prediction set to each input's difficulty. Graphs are kept or dropped by their distance to the predicted one. The authors test the approach on a synthetic task. Such promises for structured outputs could matter wherever graph forecasts guide real choices.

Catalog record

CREATORGabriel Melo, Thibaut de Saivre, Anna Calissano, Florence d’Alché-Buc
YEAR2026
COLLECTIONPAPERS
FORMATTXT
FILE SIZE43856
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2603.02460

Browse the full library · About the Tower of Babel