Title :
On learning monotone Boolean functions
Author :
Blum, Avrim ; Burch, Carl ; Langford, John
Author_Institution :
Carnegie Mellon Univ., Pittsburgh, PA, USA
Abstract :
We consider the problem of learning monotone Boolean functions over {0, 1}n under the uniform distribution. Specifically, given a polynomial number of uniform random samples for an unknown monotone Boolean function f, and given polynomial completing time, we would like to approximate f as well as possible. We describe a simple algorithm that we prove achieves error at most 1/2-Ω(1/√n), improving on the previous best bound of 1/2-Ω((log2 n)/n). We also prove that no algorithm, given a polynomial number of samples, can guarantee error 1/2-ω((log n)/√n), improving on the previous best hardness bound of O(1/√n). These lower bounds hold even if the learning algorithm is allowed membership queries. Thus this paper settles to an O(log n) factor the question of the best achievable error for learning the class of monotone Boolean functions with respect to the uniform distribution
Keywords :
Boolean functions; polynomials; lower bounds; membership queries; monotone Boolean functions learning; polynomial completing time; polynomial number; uniform random samples; Approximation algorithms; Boolean functions; Circuits; Hip; Identity-based encryption; Polynomials; Radio access networks; Upper bound;
Conference_Titel :
Foundations of Computer Science, 1998. Proceedings. 39th Annual Symposium on
Conference_Location :
Palo Alto, CA
Print_ISBN :
0-8186-9172-7
DOI :
10.1109/SFCS.1998.743491