Impossibility of Partial Recovery in the Graph Alignment Problem - Inria - Institut national de recherche en sciences et technologies du numérique Access content directly
Conference Papers Year : 2021

Impossibility of Partial Recovery in the Graph Alignment Problem

Luca Ganassali
  • Function : Author
  • PersonId : 1077335
Marc Lelarge
  • Function : Author
  • PersonId : 833445
Laurent Massoulié
  • Function : Author
  • PersonId : 933636

Abstract

Random graph alignment refers to recovering the underlying vertex correspondence between two random graphs with correlated edges. This can be viewed as an average-case and noisy version of the well-known graph isomorphism problem. For the correlated Erdős-Rényi model, we prove an impossibility result for partial recovery in the sparse regime, with constant average degree and correlation, as well as a general bound on the maximal reachable overlap. Our bound is tight in the noiseless case (the graph isomorphism problem) and we conjecture that it is still tight with noise. Our proof technique relies on a careful application of the probabilistic method to build automorphisms between tree components of a subcritical Erdős-Rényi graph.
Fichier principal
Vignette du fichier
ganassali21a.pdf (584.31 Ko) Télécharger le fichier
Origin : Files produced by the author(s)

Dates and versions

hal-03410038 , version 1 (30-10-2021)

Identifiers

  • HAL Id : hal-03410038 , version 1

Cite

Luca Ganassali, Marc Lelarge, Laurent Massoulié. Impossibility of Partial Recovery in the Graph Alignment Problem. COLT 2021 - 34th Annual Conference on Learning Theory, Aug 2021, Boulder / Virtual, United States. ⟨hal-03410038⟩
31 View
75 Download

Share

Gmail Facebook X LinkedIn More