Abstract
In this paper we prove results regarding Boolean functions with small spectral norm (the spectral norm of f is \({\|\hat{f}\|_1 = \sum_{\alpha}|\hat{f}(\alpha)|}\)). Specifically, we prove the following results for functions \({f : \{0, 1\}^n \to \{0, 1\}}\) with \({\|\hat{f}\|_1 = A}\).
-
1.
There is an affine subspace V of co-dimension at most A 2 such that \({f|_V}\) is constant.
-
2.
f can be computed by a parity decision tree of size at most \({2^{A^2} n^{2A}}\). (A parity decision tree is a decision tree whose nodes are labeled with arbitrary linear functions.)
-
3.
f can be computed by a De Morgan formula of size \({O(2^{A^2} n^{2A + 2})}\) and by a De Morgan formula of depth \({O(A^2 + \log(n) \cdot A)}\).
-
4.
If in addition f has at most s nonzero Fourier coefficients, then f can be computed by a parity decision tree of depth at most \({A^2 \log s}\).
-
5.
For every \({\epsilon > 0}\) there is a parity decision tree of depth \({O(A^2 + \log(1/\epsilon))}\) and size \({2^{O(A^2)} \cdot \min \{1/\epsilon^2, \log(1/\epsilon)^{2A}\}}\) that \({\epsilon}\)-approximates f. Furthermore, this tree can be learned (in the uniform distribution model), with probability \({1 - \delta}\), using \({{\tt poly}(n, {\rm exp}(A^2), 1/\epsilon, \log(1/\delta))}\) membership queries.
All the results above (except 3) also hold (with a slight change in parameters) for functions \({f : \mathbb{Z}_p^n \to \{0, 1\}}\).
Similar content being viewed by others
References
Alekhnovich M., Braverman M., Feldman V., Klivans A.R., Pitassi T. (2008) The complexity of properly learning simple concept classes. J. Comput. Syst. Sci. 74(1): 16–34
M. Bellare (1992). A Technique for Upper Bounding the Spectral Norm with Applications to Learning. In Proceedings of the 5th Annual COLT, 62–70.
Bellare M., Coppersmith D., Håstad J., Kiwi M.A., Sudan M. (1996) Linearity testing in characteristic two. IEEE Trans. Inf. Theory. 42(6): 1781–1795
Blum M., Luby M., Rubinfeld R. (1993) Self-testing/correcting with applications to numerical problems. J. Comput. Syst. Sci. 47(3): 549–595
Buhrman H., de Wolf R. (2002) Complexity measures and decision tree complexity: a survey. Theor. Comput. Sci. 288(1): 21–43
Friedgut E. (1998) Boolean functions with low average sensitivity depend on few coordinates. Combinatorica 18(1): 27–35
O. Goldreich & L. A. Levin (1989). A hardcore predicate for all one-way functions. In Proceedings of the 21st STOC, 25–32.
Gopalan P., O’Donnell R., Servedio R., Shpilka A., Wimmer K. (2011) Testing Fourier dimensionality and sparsity. SIAM J. Comput. 40(4): 1075–1100
Green B., Sanders T. (2008a) Boolean functions with small spectral norm. GAFA 18: 144–162
Green B., Sanders T. (2008b) A quantitative version of the idempotent theorem in harmonic analysis. Annals of Math. 168(3): 1025–1054
Grolmusz V. (1997) On the power of circuits with gates of low L 1 norms. Theor. Comput. Sci. 188(1): 117–128
Håstad J. (2001) Some optimal inapproximability results. J. ACM 48(4): 798–859
J. Kahn, G. Kalai & N. Linial (1988). The influence of variables on Boolean functions. In Proceedings of the 29th annual FOCS, 68–80.
Kalai G. (2002) A Fourier-theoretic perspective on the Condorcet paradox and Arrow’s theorem. Adv. Appl. Math. 29(3): 412–426
Kushilevitz E., Mansour Y. (1993) Learning Decision Trees Using the Fourier Spectrum. SIAM J. Comput. 22(6): 1331–1348
Linial N., Mansour Y., Nisan N. (1993) Constant Depth Circuits, Fourier Transform and Learnability. J. ACM 40(3): 607–620
S. Lovett (2014). Communication is bounded by root of rank. In Proceedings of the 46th Annual ACM Symposium on Theory of Computing (STOC).
Y. Mansour (1994). Learning Boolean functions via the Fourier transform. In Theoretical advances in neural computation and learning, 391–424.
A. Montanaro & T. Osborne (2009). On the communication complexity of XOR functions. CoRR abs/0909.3392.
R. O’Donnell (2014). Analysis of boolean functions. Cambridge University Press.
A. Shpilka, A. Tal & B. L. Volk (2014). On the structure of boolean functions with small spectral norm. In Innovations in Theoretical Computer Science, ITCS’14, 37–48.
P. M. Spira (1971). On time-hardware complexity tradeoffs for Boolean functions. In Proceedings of the 4th HICSS, 525–527.
H. Y. Tsang, C. H. Wong, N. Xie & S. Zhang (2013). Fourier sparsity, spectral norm, and the Log-rank conjecture. In Proceedings of the 54th Annual FOCS, 658–667.
S. V. Yablonskii (1954). Realization of the linear function in the class of \({\Pi}\)-schemes. In Dokl. Akad. Nauk SSSR, volume 94, 805–806. In Russian.
Z. Zhang & Y. Shi (2010). On the parity complexity measures of Boolean functions. Theor. Comput. Sci. 411(26-28), 2612 – 2618. ISSN 0304-3975.
Author information
Authors and Affiliations
Corresponding author
Rights and permissions
About this article
Cite this article
Shpilka, A., Tal, A. & Volk, B.l. On the Structure of Boolean Functions with Small Spectral Norm. comput. complex. 26, 229–273 (2017). https://doi.org/10.1007/s00037-015-0110-y
Received:
Published:
Issue Date:
DOI: https://doi.org/10.1007/s00037-015-0110-y