Nothing Special   »   [go: up one dir, main page]

skip to main content
10.1145/3313276.3316360acmconferencesArticle/Chapter ViewAbstractPublication PagesstocConference Proceedingsconference-collections
research-article

Reconstruction of non-degenerate homogeneous depth three circuits

Published: 23 June 2019 Publication History

Abstract

A homogeneous depth three circuit C computes a polynomial f = T1 + T2 + ... + Ts, where each Ti is a product of d linear forms in n variables over some underlying field F. Given black-box access to f, can we efficiently reconstruct (i.e. proper learn) a homogeneous depth three circuit computing f? Learning various subclasses of circuits is natural and interesting from both theoretical and practical standpoints and in particular, properly learning homogeneous depth three circuits efficiently is stated as an open problem in a work by Klivans and Shpilka (COLT 2003) and is well-studied. Unfortunately, there is substantial amount of evidence to show that this is a hard problem in the worst case. We give a (randomized) poly(n,d,s)-time algorithm to reconstruct non-degenerate homogeneous depth three circuits for n = Ω(d2) (with some additional mild requirements on s and the characteristic of F). We call a circuit C as non-degenerate if the dimension of the partial derivative space of f equals the sum of the dimensions of the partial derivative spaces of the terms T1, T2, …, Ts. In this sense, the terms are “independent” of each other in a non-degenerate circuit. A random homogeneous depth three circuit (where the coefficients of the linear forms are chosen according to the uniform distribution or any other reasonable distribution) is almost surely non-degenerate. In comparison, previous learning algorithms for this circuit class were either improper (with an exponential dependence on d), or they only worked for s < n (with a doubly exponential dependence of the running time on s). The main contribution of this work is to formulate the following paradigm for efficiently handling addition gates and to successfully implement it for the class of homogeneous depth three circuits. The problem of finding the children of an addition gate with large fan-in s is first reduced to the problem of decomposing a suitable vector space U into a (direct) sum of simpler subspaces U1, U2, …, Us. One then constructs a suitable space of operators S consisting of linear maps acting on U such that analyzing the simultaneous global structure of S enables us to efficiently decompose U. In our case, we exploit the structure of the set of low rank matrices in S and of the invariant subspaces of U induced by S. We feel that this paradigm is novel and powerful: it should lead to efficient reconstruction of many other subclasses of circuits for which the efficient reconstruction problem had hitherto looked unapproachable because of the presence of large fan-in addition gates.

References

[1]
Manindra Agrawal and V. Vinay. 2008. Arithmetic Circuits: A Chasm at Depth Four. In 49th Annual IEEE Symposium on Foundations of Computer Science, FOCS 2008, October 25-28, 2008, Philadelphia, PA, USA. 67–75.
[2]
Dana Angluin. 1988.
[3]
Queries and Concept Learning. Machine Learning. 2, 4 (1988), 319–342.
[4]
Nikhil Balaji, Nutan Limaye, and Srikanth Srinivasan. 2016.
[5]
An Almost Cubic Lower Bound for ΣΠΣ Circuits Computing a Polynomial in VP. Electronic Colloquium on Computational Complexity (ECCC) 23 (2016), 143.
[6]
Amos Beimel, Francesco Bergadano, Nader H. Bshouty, Eyal Kushilevitz, and Stefano Varricchio. 2000. Learning functions represented as multiplicity automata. J. ACM 47, 3 (2000), 506–530.
[7]
Michael Ben-Or and Richard Cleve. 1992. Computing Algebraic Formulas Using a Constant Number of Registers. SIAM J. Comput. 21, 1 (1992), 54–58.
[8]
Michael Ben-Or and Prasoon Tiwari. 1988. A Deterministic Algorithm for Sparse Multivariate Polynominal Interpolation (Extended Abstract). In Proceedings of the 20th Annual ACM Symposium on Theory of Computing, May 2-4, 1988, Chicago, Illinois, USA. 301–309.
[9]
Elwyn R Berlekamp. 1970. Factoring polynomials over large finite fields. Math. Comp. 24 (1970), 713–735.
[10]
Marco L. Carmosino, Russell Impagliazzo, Valentine Kabanets, and Antonina Kolokolova. 2016. Learning Algorithms from Natural Proofs. In 31st Conference on Computational Complexity, CCC 2016, May 29 to June 1, 2016, Tokyo, Japan. 10:1–10:24.
[11]
Michael A. Forbes, Amir Shpilka, and Ben Lee Volk. 2017. Succinct hitting sets and barriers to proving algebraic circuits lower bounds. In Proceedings of the 49th Annual ACM SIGACT Symposium on Theory of Computing, STOC 2017, Montreal, QC, Canada, June 19-23, 2017. 653–664.
[12]
Lance Fortnow and Adam R. Klivans. 2009. Efficient learning algorithms yield circuit lower bounds. J. Comput. Syst. Sci. 75, 1 (2009), 27–36.
[13]
Ignacio García-Marco, Pascal Koiran, and Timothée Pecatte. 2018. Polynomial Equivalence Problems for Sum of Affine Powers. In Proceedings of the 2018 ACM on International Symposium on Symbolic and Algebraic Computation, ISSAC 2018, New York, NY, USA, July 16-19, 2018. 303–310.
[14]
Dima Grigoriev, Marek Karpinski, and Michael F. Singer. 1994. Computational Complexity of Sparse Rational Interpolation. SIAM J. Comput. 23, 1 (1994), 1–11.
[15]
Joshua A. Grochow, Mrinal Kumar, Michael E. Saks, and Shubhangi Saraf. 2017.
[16]
Towards an algebraic natural proofs barrier via polynomial identity testing. CoRR abs/1701.01717 (2017).
[17]
Ankit Gupta, Pritish Kamath, Neeraj Kayal, and Ramprasad Saptharishi. 2016.
[18]
Arithmetic Circuits: A Chasm at Depth 3. SIAM J. Comput. 45, 3 (2016), 1064–1079.
[19]
Johan Håstad. 1990.
[20]
Tensor Rank is NP-Complete. J. Algorithms 11, 4 (1990), 644–654.
[21]
Jeffrey C. Jackson, Adam R. Klivans, and Rocco A. Servedio. 2002.
[22]
Learnability beyond AC0. In Proceedings on 34th Annual ACM Symposium on Theory of Computing, May 19-21, 2002, Montréal, Québec, Canada. 776–784.
[23]
Erich Kaltofen and Barry M. Trager. 1990. Computing with Polynomials Given By Black Boxes for Their Evaluations: Greatest Common Divisors, Factorization, Separation of Numerators and Denominators. J. Symb. Comput. 9, 3 (1990), 301–320.
[24]
Zohar Shay Karnin and Amir Shpilka. 2009. Reconstruction of Generalized Depth- 3 Arithmetic Circuits with Bounded Top Fan-in. In Proceedings of the 24th Annual IEEE Conference on Computational Complexity, CCC 2009, Paris, France, 15-18 July 2009. 274–285.
[25]
Neeraj Kayal. 2011. Efficient algorithms for some special cases of the polynomial equivalence problem. In Proceedings of the Twenty-Second Annual ACM-SIAM Symposium on Discrete Algorithms, SODA 2011, San Francisco, California, USA, January 23-25, 2011. 1409–1421.
[26]
Neeraj Kayal. 2012.
[27]
Affine projections of polynomials: extended abstract. In Proceedings of the 44th Symposium on Theory of Computing Conference, STOC 2012, New York, NY, USA, May 19 - 22, 2012. 643–662.
[28]
Neeraj Kayal, Nutan Limaye, Chandan Saha, and Srikanth Srinivasan. 2017. An Exponential Lower Bound for Homogeneous Depth Four Arithmetic Formulas. SIAM J. Comput. 46, 1 (2017), 307–335.
[29]
Neeraj Kayal, Vineet Nair, and Chandan Saha. 2018. Average-case linear matrix factorization and reconstruction of low width Algebraic Branching Programs. Electronic Colloquium on Computational Complexity (ECCC) 25 (2018), 29.
[30]
Neeraj Kayal, Vineet Nair, Chandan Saha, and Sébastien Tavenas. 2017. Reconstruction of Full Rank Algebraic Branching Programs. In 32nd Computational Complexity Conference, CCC 2017, July 6-9, 2017, Riga, Latvia. 21:1–21:61.
[31]
Neeraj Kayal, Chandan Saha, and Ramprasad Saptharishi. 2014.
[32]
A superpolynomial lower bound for regular arithmetic formulas. In Symposium on Theory of Computing, STOC 2014, New York, NY, USA, May 31 - June 03, 2014. 146–153.
[33]
Neeraj Kayal, Chandan Saha, and Sébastien Tavenas. 2016.
[34]
An Almost Cubic Lower Bound for Depth Three Arithmetic Circuits. In 43rd International Colloquium on Automata, Languages, and Programming, ICALP 2016, July 11-15, 2016, Rome, Italy. 33:1–33:15.
[35]
Adam R. Klivans and Amir Shpilka. 2003. Learning Arithmetic Circuits via Partial Derivatives. In 16th Annual Conference on Computational Learning Theory (COLT), Washington, DC, USA, August 24-27, 2003, Proceedings. 463–476.
[36]
Adam R. Klivans and Daniel A. Spielman. 2001. Randomness efficient identity testing of multivariate polynomials. In Proceedings on 33rd Annual ACM Symposium on Theory of Computing, July 6-8, 2001, Heraklion, Crete, Greece. 216–223.
[37]
Pascal Koiran. 2012.
[38]
Arithmetic circuits: The chasm at depth four gets wider. Theor. Comput. Sci. 448 (2012), 56–65.
[39]
Mrinal Kumar. 2017.
[40]
A Quadratic Lower Bound for Homogeneous Algebraic Branching Programs. In Proceedings of the 32nd Computational Complexity Conference (CCC ’17). Article 19, 16 pages.
[41]
Arjen K Lenstra, Hendrik W Lenstra, and László Lovász. 1982. Factoring polynomials with rational coefficients. Math. Ann. 261, 4 (1982), 515–534.
[42]
Nathan Linial, Yishay Mansour, and Noam Nisan. 1993. Constant Depth Circuits, Fourier Transform, and Learnability. J. ACM 40, 3 (1993), 607–620.
[43]
Noam Nisan and Avi Wigderson. 1997. Lower Bounds on Arithmetic Circuits Via Partial Derivatives. Computational Complexity 6, 3 (1997), 217–234.
[44]
Alexander A. Razborov and Steven Rudich. 1997. Natural Proofs. J. Comput. Syst. Sci. 55, 1 (1997), 24–35.
[45]
Jacob T. Schwartz. 1980. Fast Probabilistic Algorithms for Verification of Polynomial Identities. J. ACM 27, 4 (1980), 701–717.
[46]
Yaroslav Shitov. 2016. How hard is the tensor rank? arXiv abs/1611.01559 (2016).
[47]
https://arxiv.org/abs/1611.01559
[48]
Amir Shpilka. 2007. Interpolation of depth-3 arithmetic circuits with two multiplication gates. In Proceedings of the 39th Annual ACM Symposium on Theory of Computing, San Diego, California, USA, June 11-13, 2007. 284–293.
[49]
Amir Shpilka and Avi Wigderson. 2001. Depth-3 arithmetic circuits over fields of characteristic zero. Computational Complexity 10, 1 (2001), 1–27.
[50]
Gaurav Sinha. 2016. Reconstruction of Real Depth-3 Circuits with Top Fan-In 2. In 31st Conference on Computational Complexity, CCC 2016, May 29 to June 1, 2016, Tokyo, Japan. 31:1–31:53.
[51]
Sébastien Tavenas. 2013. Improved Bounds for Reduction to Depth 4 and Depth 3. In Mathematical Foundations of Computer Science 2013 - 38th International Symposium, MFCS 2013, Klosterneuburg, Austria, August 26-30, 2013. Proceedings. 813–824.
[52]
Ilya Volkovich. 2016. A Guide to Learning Arithmetic Circuits. In Proceedings of the 29th Conference on Learning Theory, COLT 2016, New York, USA, June 23-26, 2016. 1540–1561.
[53]
Avi Wigderson. 2007.
[54]
P, NP and Mathematics - a computational complexity perspective. In Proceedings of the International Congress of Mathematicians (ICM), 2006, Madrid, Spain. 665–712.
[55]
Morris Yau. 2016. Almost Cubic Bound for Depth Three Circuits in VP. Electronic Colloquium on Computational Complexity (ECCC) 23 (2016), 187.

Cited By

View all
  • (2023)Absolute reconstruction for sums of powers of linear forms: degree 3 and beyondcomputational complexity10.1007/s00037-023-00239-832:2Online publication date: 3-Aug-2023
  • (2022)Polynomial-Time Power-Sum Decomposition of Polynomials2022 IEEE 63rd Annual Symposium on Foundations of Computer Science (FOCS)10.1109/FOCS54457.2022.00094(956-967)Online publication date: Oct-2022
  • (2021)Hitting sets and reconstruction for dense orbits in VPe and ΣΠΣ circuitsProceedings of the 36th Computational Complexity Conference10.4230/LIPIcs.CCC.2021.19Online publication date: 20-Jul-2021
  • Show More Cited By

Index Terms

  1. Reconstruction of non-degenerate homogeneous depth three circuits

    Recommendations

    Comments

    Please enable JavaScript to view thecomments powered by Disqus.

    Information & Contributors

    Information

    Published In

    cover image ACM Conferences
    STOC 2019: Proceedings of the 51st Annual ACM SIGACT Symposium on Theory of Computing
    June 2019
    1258 pages
    ISBN:9781450367059
    DOI:10.1145/3313276
    Permission to make digital or hard copies of all or part of this work for personal or classroom use is granted without fee provided that copies are not made or distributed for profit or commercial advantage and that copies bear this notice and the full citation on the first page. Copyrights for components of this work owned by others than the author(s) must be honored. Abstracting with credit is permitted. To copy otherwise, or republish, to post on servers or to redistribute to lists, requires prior specific permission and/or a fee. Request permissions from [email protected].

    Sponsors

    Publisher

    Association for Computing Machinery

    New York, NY, United States

    Publication History

    Published: 23 June 2019

    Permissions

    Request permissions for this article.

    Check for updates

    Author Tags

    1. Circuit reconstruction
    2. homogeneous depth three circuits
    3. invariant subspaces
    4. shifted differential operators

    Qualifiers

    • Research-article

    Conference

    STOC '19
    Sponsor:

    Acceptance Rates

    Overall Acceptance Rate 1,469 of 4,586 submissions, 32%

    Contributors

    Other Metrics

    Bibliometrics & Citations

    Bibliometrics

    Article Metrics

    • Downloads (Last 12 months)18
    • Downloads (Last 6 weeks)4
    Reflects downloads up to 19 Nov 2024

    Other Metrics

    Citations

    Cited By

    View all
    • (2023)Absolute reconstruction for sums of powers of linear forms: degree 3 and beyondcomputational complexity10.1007/s00037-023-00239-832:2Online publication date: 3-Aug-2023
    • (2022)Polynomial-Time Power-Sum Decomposition of Polynomials2022 IEEE 63rd Annual Symposium on Foundations of Computer Science (FOCS)10.1109/FOCS54457.2022.00094(956-967)Online publication date: Oct-2022
    • (2021)Hitting sets and reconstruction for dense orbits in VPe and ΣΠΣ circuitsProceedings of the 36th Computational Complexity Conference10.4230/LIPIcs.CCC.2021.19Online publication date: 20-Jul-2021
    • (2021)Reconstruction algorithms for low-rank tensors and depth-3 multilinear circuitsProceedings of the 53rd Annual ACM SIGACT Symposium on Theory of Computing10.1145/3406325.3451096(809-822)Online publication date: 15-Jun-2021
    • (2021)Derandomization and Absolute Reconstruction for Sums of Powers of Linear FormsTheoretical Computer Science10.1016/j.tcs.2021.07.005Online publication date: Jul-2021
    • (2020)Implementing geometric complexity theory: on the separation of orbit closures via symmetriesProceedings of the 52nd Annual ACM SIGACT Symposium on Theory of Computing10.1145/3357713.3384257(713-726)Online publication date: 22-Jun-2020
    • (2020)Learning sums of powers of low-degree polynomials in the non-degenerate case2020 IEEE 61st Annual Symposium on Foundations of Computer Science (FOCS)10.1109/FOCS46700.2020.00087(889-899)Online publication date: Nov-2020

    View Options

    Login options

    View options

    PDF

    View or Download as a PDF file.

    PDF

    eReader

    View online with eReader.

    eReader

    Media

    Figures

    Other

    Tables

    Share

    Share

    Share this Publication link

    Share on social media