On the Fourier spectrum of symmetric Boolean functions

We study the following questionWhat is the smallest t such that every symmetric boolean function on κ variables (which is not a constant or a parity function), has a non-zero Fourier coefficient of order at least 1 and at most t?We exclude the constant functions for which there is no such t and the parity functions for which t has to be κ. Let τ (κ) be the smallest such t. Our main result is that for large κ, τ (κ)≤4κ/logκ.The motivation for our work is to understand the complexity of learning symmetric juntas. A κ-junta is a boolean function of n variables that depends only on an unknown subset of κ variables. A symmetric κ-junta is a junta that is symmetric in the variables it depends on. Our result implies an algorithm to learn the class of symmetric κ-juntas, in the uniform PAC learning model, in time no(κ). This improves on a result of Mossel, O’Donnell and Servedio in [16], who show that symmetric κ-juntas can be learned in time n2κ/3.

[1]  Avrim Blum,et al.  Relevant Examples and Relevant Features: Thoughts from Computational Learning Theory , 1994 .

[2]  Karsten A. Verbeurgt.  Learning Sub-classes of Monotone DNF on the Uniform Distribution , 1998, ALT.

[3]  Jeffrey C. Jackson,et al.  An efficient membership-query algorithm for learning DNF with respect to the uniform distribution , 1994, Proceedings 35th Annual Symposium on Foundations of Computer Science.

[4]  P. Cameron.  Combinatorics: Topics, Techniques, Algorithms , 1995 .

[5]  Ryan O'Donnell,et al.  Learning juntas , 2003, STOC '03.

[6]  Nader H. Bshouty,et al.  More efficient PAC-learning of DNF with membership queries under the uniform distribution , 2004, J. Comput. Syst. Sci..

[7]  Joachim von zur Gathen,et al.  Polynomials with two values , 1997, Comb..

[8]  Thomas Siegenthaler,et al.  Correlation-immunity of nonlinear combining functions for cryptographic applications , 1984, IEEE Trans. Inf. Theory.

[9]  G. Pólya,et al.  Problems and theorems in analysis , 1983 .

[10]  Karsten A. Verbeurgt.  Learning DNF under the uniform distribution in quasi-polynomial time , 1990, COLT '90.

[11]  Pat Langley,et al.  Selection of Relevant Features and Examples in Machine Learning , 1997, Artif. Intell..

[12]  Nisheeth K. Vishnoi,et al.  On the Fourier spectrum of symmetric Boolean functions with applications to learning symmetric juntas , 2005, 20th Annual IEEE Conference on Computational Complexity (CCC'05).

[13]  Noga Alon,et al.  Testing k-wise and almost k-wise independence , 2007, STOC '07.

[14]  Noam Nisan,et al.  Constant depth circuits, Fourier transform, and learnability , 1993, JACM.

[15]  Yishay Mansour,et al.  An O(nlog log n) learning algorithm for DNF under the uniform distribution , 1992, COLT '92.

[16]  Aranyak Mehta,et al.  Learning symmetric k-juntas in time n , 2005 .

[17]  Angel V. Kumchev,et al.  The Distribution of Prime Numbers , 2005 .

[18]  Leslie G. Valiant,et al.  A theory of the learnable , 1984, CACM.

[19]  Richard J. Lipton,et al.  Cryptographic Primitives Based on Hard Learning Problems , 1993, CRYPTO.

[20]  Klaas Pieter Hart,et al.  Open Problems , 2022, Dimension Groups and Dynamical Systems.

[21]  Aranyak Mehta,et al.  Learning symmetric k-juntas in time n o(k) , 2005 .

[22]  Manfred K. Warmuth,et al.  Learning integer lattices , 1990, COLT '90.