Learning Unions of ω(1)-Dimensional Rectangles

  • Alp Atıcı
  • Rocco A. Servedio
Conference paper
Part of the Lecture Notes in Computer Science book series (LNCS, volume 4264)


We consider the problem of learning unions of rectangles over the domain [b] n , in the uniform distribution membership query learning setting, where both b and n are “large”. We obtain poly(n, logb)-time algorithms for the following classes:

– poly (n logb)-Majority of \(O(\frac{\log(n \log b)} {\log \log(n \log b)})\)-dimensional rectangles.

–Unions of poly(log(n logb)) many rectangles with dimension

\(O(\frac{\log^2 (n \log b)} {(\log \log(n \log b) \log \log \log (n \log b))^2})\).

– poly (n logb)-Majority of poly (n logb)-Or of disjoint rectangles

with dimension \(O(\frac{\log(n \log b)} {\log \log(n \log b)})\)

Our main algorithmic tool is an extension of Jackson’s boosting- and Fourier-based Harmonic Sieve algorithm [13] to the domain [b] n , building on work of Akavia et al. [1]. Other ingredients used to obtain the results stated above are techniques from exact learning [4] and ideas from recent work on learning augmented AC 0 circuits [14] and on representing Boolean functions as thresholds of parities [16].


Boolean Function Concept Class Weak Hypothesis Membership Query Equivalence Query 
These keywords were added by machine and not by the authors. This process is experimental and the keywords may be updated as the learning algorithm improves.


Unable to display preview. Download preview PDF.

Unable to display preview. Download preview PDF.


  1. 1.
    Akavia, A., Goldwasser, S., Safra, S.: Proving Hard Core Predicates Using List Decoding. In: Proc. 44th IEEE Found. Comp. Sci., pp. 146–156 (2003)Google Scholar
  2. 2.
    Aizenstein, H., Blum, A., Khardon, R., Kushilevitz, E., Pitt, L., Roth, D.: On Learning Read-k Satisfy-j DNF. SIAM Journal on Computing 27(6), 1515–1530 (1998)MATHCrossRefMathSciNetGoogle Scholar
  3. 3.
    Atıcı, A., Servedio, R.: Learning Unions of ω(1)-Dimensional Rectangles, available at:
  4. 4.
    Beimel, A., Kushilevitz, E.: Learning Boxes in High Dimension. Algorithmica 22(1/2), 76–90 (1998)MATHCrossRefMathSciNetGoogle Scholar
  5. 5.
    Bruck, J.: Harmonic Analysis of Polynomial Threshold Functions. SIAM Journal on Discrete Mathematics 3(2), 168–177 (1990)MATHCrossRefMathSciNetGoogle Scholar
  6. 6.
    Chen, Z., Homer, S.: The Bounded Injury Priority Method and The Learnability of Unions of Rectangles. Annals of Pure and Applied Logic 77(2), 143–168 (1996)MATHCrossRefMathSciNetGoogle Scholar
  7. 7.
    Chen, Z., Maass, W.: On-line Learning of Rectangles and Unions of Rectangles. Machine Learning 17(2/3), 23–50 (1994)CrossRefGoogle Scholar
  8. 8.
    Freund, Y.: Boosting a Weak Learning Algorithm by Majority. Information and Computation 121(2), 256–285 (1995)MATHCrossRefMathSciNetGoogle Scholar
  9. 9.
    Freund, Y., Schapire, R.: A Short Introduction to Boosting. Journal of the Japanese Society for Artificial Intelligence 14(5), 771–780 (1999)Google Scholar
  10. 10.
    Goldberg, P.W., Goldman, S.A., Mathias, H.D.: Learning Unions of Boxes with Membership and Equivalence Queries. In: COLT 1994: Proc. of the 7th annual conference on computational learning theory, pp. 198–207 (1994)Google Scholar
  11. 11.
    Hajnal, A., Maass, W., Pudlák, P., Szegedy, M., Turan, G.: Threshold Circuits of Bounded Depth. J. Comp. & Syst. Sci. 46, 129–154 (1993)MATHCrossRefGoogle Scholar
  12. 12.
    Håstad, J.: Computational Limitations for Small Depth Circuits. MIT Press, Cambridge, MA (1986)Google Scholar
  13. 13.
    Jackson, J.C.: An Efficient Membership-Query Algorithm for Learning DNF with Respect to the Uniform Distribution. J. Comp. & Syst. Sci. 55(3), 414–440 (1997)MATHCrossRefGoogle Scholar
  14. 14.
    Jackson, J.C., Klivans, A.R., Servedio, R.A.: Learnability Beyond AC 0. In: Proc. of the 34th annual ACM symposium on theory of computing (STOC), pp. 776–784 (2002)Google Scholar
  15. 15.
    Khardon, R.: On Using the Fourier Transform to Learn Disjoint DNF. Information Processing Letters 49(5), 219–222 (1994)MATHCrossRefGoogle Scholar
  16. 16.
    Klivans, A.R., Servedio, R.A.: Learning DNF in Time \(2^{\tilde{O}(n^{1/3})}\). J. Comp. & Syst. Sci. 68(2), 303–318 (2004)MATHCrossRefMathSciNetGoogle Scholar
  17. 17.
    Krause, M., Pudlák, P.: Computing Boolean Functions by Polynomials and Threshold Circuits. Computational Complexity 7(4), 346–370 (1998)MATHCrossRefMathSciNetGoogle Scholar
  18. 18.
    Kushilevitz, E., Mansour, Y.: Learning Decision Trees using the Fourier Spectrum. SIAM Journal on Computing 22(6), 1331–1348 (1993)MATHCrossRefMathSciNetGoogle Scholar
  19. 19.
    Maass, W., Warmuth, M.K.: Efficient Learning with Virtual Threshold Gates. Information and Computation 141(1), 66–83 (1998)MATHCrossRefMathSciNetGoogle Scholar
  20. 20.
    Schapire, R.E.: The Strength of Weak Learnability. Machine Learning 5, 197–227 (1990)Google Scholar

Copyright information

© Springer-Verlag Berlin Heidelberg 2006

Authors and Affiliations

  • Alp Atıcı
    • 1
  • Rocco A. Servedio
    • 1
  1. 1.Columbia UniversityNew YorkUSA

Personalised recommendations