On Exact Learning Monotone DNF from Membership Queries

TL;DR

This paper investigates the problem of exactly learning monotone DNF formulas with bounded size and number of terms using membership queries, providing new bounds and efficient algorithms with near-optimal query complexity.

Contribution

It introduces new lower bounds and develops deterministic and randomized adaptive algorithms that are nearly optimal in query complexity for learning monotone DNF.

Findings

New lower bounds for learning monotone DNF from membership queries.

Deterministic and randomized algorithms with near-optimal query complexity.

Algorithms run in linear time relative to query complexity and number of variables.

Abstract

In this paper, we study the problem of learning a monotone DNF with at most $s$ terms of size (number of variables in each term) at most $r$ ($s$ term $r$-MDNF) from membership queries. This problem is equivalent to the problem of learning a general hypergraph using hyperedge-detecting queries, a problem motivated by applications arising in chemical reactions and genome sequencing. We first present new lower bounds for this problem and then present deterministic and randomized adaptive algorithms with query complexities that are almost optimal. All the algorithms we present in this paper run in time linear in the query complexity and the number of variables $n$. In addition, all of the algorithms we present in this paper are asymptotically tight for fixed $r$ and/or $s$.