Abstract
In 2002 Jackson [JKS02] asked whether AC 0 circuits augmented with a threshold gate at the output can be efficiently learned from uniform random examples. We answer this question affirmatively by showing that such circuits have fairly strong Fourier concentration; hence the low-degree algorithm of Linial, Mansour and Nisan [LMN93] learns such circuits in sub-exponential time. Under a conjecture of Gotsman and Linial [GL94] which upper bounds the total influence of low-degree polynomial threshold functions, the running time is quasi-polynomial. Our results extend to AC 0 circuits augmented with a small super-constant number of threshold gates at arbitrary locations in the circuit. We also establish some new structural properties of AC 0 circuits augmented with threshold gates, which allow us to prove a range of separation results and lower bounds.
Access this chapter
Tax calculation will be finalised at checkout
Purchases are for personal use only
Preview
Unable to display preview. Download preview PDF.
References
Aspnes, J., Beigel, R., Furst, M., Rudich, S.: The expressive power of voting polynomials. Combinatorica 14(2), 1–14 (1994)
Beigel, R.: When do extra majority gates help? polylog(n) majority gates are equivalent to one. Computational Complexity 4, 314–324 (1994)
Blais, E., O’Donnell, R., Wimmer, K.: Polynomial regression under arbitrary product distributions. In: COLT, pp. 193–204 (2008)
Beigel, R., Reingold, N., Spielman, D.: PP is closed under intersection. Journal of Computer & System Sciences 50(2), 191–202 (1995)
Bshouty, N., Tamon, C.: On the Fourier spectrum of monotone functions. Journal of the ACM 43(4), 747–770 (1996)
Diakonikolas, I., Lee, H., Matulef, K., Onak, K., Rubinfeld, R., Servedio, R., Wan, A.: Testing for concise representations. In: FOCS, pp. 549–558 (2007)
Diakonikolas, I., Raghavendra, P., Servedio, R., Tan, L.-Y.: Average sensitivity and noise sensitivity of polynomial threshold functions (2009), http://arxiv.org/abs/0909.5011
Furst, M., Jackson, J., Smith, S.: Improved learning of AC 0 functions. In: COLT, pp. 317–325 (1991)
Fortnow, L., Klivans, A.: Efficient learning algorithms yield circuit lower bounds. Journal of Computer & System Sciences 75(1), 27–36 (2009)
Goldmann, M., Håstad, J., Razborov, A.: Majority gates vs. general weighted threshold gates. Computational Complexity 2, 277–300 (1992)
Gotsman, C., Linial, N.: Spectral properties of threshold functions. Combinatorica 14(1), 35–50 (1994)
Goldmann, M.: On the power of a threshold gate at the top. Information Processing Letters 63(6), 287–293 (1997)
Gopalan, P., Servedio, R.A.: Learning and Lower Bounds for AC 0 with Threshold Gates (2010), http://eccc.hpi-web.de/report/2010/074/
Hansen, K.: Computing symmetric Boolean functions by circuits with few exact threshold gates. In: COCOON, pp. 448–458 (2007)
HÃ¥stad, J.: Computational Limitations for Small Depth Circuits. MIT Press, Cambridge (1986)
Håstad, J.: A slight sharpening of LMN. Journal of Computer and System Sciences 63(3), 498–508 (2001)
Harsha, P., Klivans, A., Meka, R.: Bounding the sensitivity of polynomial threshold functions (2009), http://arxiv.org/abs/0909.5175
Jackson, J., Klivans, A., Servedio, R.: Learnability beyond AC 0. In: STOC, pp. 776–784 (2002)
Kalai, A., Klivans, A., Mansour, Y., Servedio, R.: Agnostically learning halfspaces. SIAM Journal on Computing 37(6), 1777–1805 (2008)
Klivans, A., O’Donnell, R., Servedio, R.: Learning intersections and thresholds of halfspaces. Journal of Computer & System Sciences 68(4), 808–840 (2004)
Klivans, A., O’Donnell, R., Servedio, R.: Learning geometric concepts via Gaussian surface area. In: FOCS, pp. 541–550 (2008)
Linial, N., Mansour, Y., Nisan, N.: Constant depth circuits, Fourier transform and learnability. Journal of the ACM 40(3), 607–620 (1993)
Mossel, E., O’Donnell, R., Servedio, R.: Learning functions of k relevant variables. Journal of Computer & System Sciences 69(3), 421–434 (2004)
O’Donnell, R., Servedio, R.: Learning monotone decision trees in polynomial time. SIAM J. Comput. 37(3), 827–844 (2007)
Podolskii, V.: Personal communication (2010)
Sherstov, A.: The intersection of two halfspaces has high threshold degree. In: FOCS, pp. 343–362 (2009)
Author information
Authors and Affiliations
Editor information
Editors and Affiliations
Rights and permissions
Copyright information
© 2010 Springer-Verlag Berlin Heidelberg
About this paper
Cite this paper
Gopalan, P., Servedio, R.A. (2010). Learning and Lower Bounds for AC 0 with Threshold Gates. In: Serna, M., Shaltiel, R., Jansen, K., Rolim, J. (eds) Approximation, Randomization, and Combinatorial Optimization. Algorithms and Techniques. RANDOM APPROX 2010 2010. Lecture Notes in Computer Science, vol 6302. Springer, Berlin, Heidelberg. https://doi.org/10.1007/978-3-642-15369-3_44
Download citation
DOI: https://doi.org/10.1007/978-3-642-15369-3_44
Publisher Name: Springer, Berlin, Heidelberg
Print ISBN: 978-3-642-15368-6
Online ISBN: 978-3-642-15369-3
eBook Packages: Computer ScienceComputer Science (R0)