English

On Multistage Learning a Hidden Hypergraph

Information Theory 2016-11-18 v3 math.IT

Abstract

Learning a hidden hypergraph is a natural generalization of the classical group testing problem that consists in detecting unknown hypergraph Hun=H(V,E)H_{un}=H(V,E) by carrying out edge-detecting tests. In the given paper we focus our attention only on a specific family F(t,s,)F(t,s,\ell) of localized hypergraphs for which the total number of vertices V=t|V| = t, the number of edges Es|E|\le s, sts\ll t, and the cardinality of any edge e|e|\le\ell, t\ell\ll t. Our goal is to identify all edges of HunF(t,s,)H_{un}\in F(t,s,\ell) by using the minimal number of tests. We develop an adaptive algorithm that matches the information theory bound, i.e., the total number of tests of the algorithm in the worst case is at most slog2t(1+o(1))s\ell\log_2 t(1+o(1)). We also discuss a probabilistic generalization of the problem.

Keywords

Cite

@article{arxiv.1601.06705,
  title  = {On Multistage Learning a Hidden Hypergraph},
  author = {A. G. D'yachkov and I. V. Vorobyev and N. A. Polyanskii and V. Yu. Shchukin},
  journal= {arXiv preprint arXiv:1601.06705},
  year   = {2016}
}

Comments

5 pages, IEEE conference

R2 v1 2026-06-22T12:36:14.867Z