Abstract
The perceptron algorithm for linear programming, arising from machine learning, has been around since the 1950s. While not a polynomial-time algorithm, it is useful in practice due to its simplicity and robustness. In 2004, Dunagan and Vempala showed that a randomized rescaling turns the perceptron method into a polynomial time algorithm, and later Peña and Soheili gave a deterministic rescaling. In this paper, we give a deterministic rescaling for the perceptron algorithm that improves upon the previous rescaling methods by making it possible to rescale much earlier. This results in a faster running time for the rescaled perceptron algorithm. We will also demonstrate that the same rescaling methods yield a polynomial time algorithm based on the multiplicative weights update method. This draws a connection to an area that has received a lot of recent attention in theoretical computer science.
T. Rothvoss—Supported by an Alfred P. Sloan Research Fellowship. Both authors supported by NSF grant 1420180 with title “Limitations of convex relaxations in combinatorial optimization”.
Access this chapter
Tax calculation will be finalised at checkout
Purchases are for personal use only
Notes
- 1.
The \(\tilde{O}\)-notation suppresses any \({\text {polylog}}(m,n)\) terms.
- 2.
It suffices here to consider the trivial example with \(\lambda _1=\ldots = \lambda _n = \frac{1}{n}\) and \(A_i = e_i\) being the standard basis. Then \(\Vert \sum _{i \in J} \lambda _i A_i\Vert _2 \le \frac{1}{\sqrt{n}}\) for any subset J. The optimality of our rescaling can also be seen since the cone in the last iteration is \(\tilde{O}(n)\)-well rounded, which is optimal up to \(\tilde{O}\)-terms.
- 3.
Recall that a function \(F : {\mathbb {R}}^n \rightarrow {\mathbb {R}}\) is Lipschitz with Lipschitz constant 1 if \(|F(x)-F(y)| \le \Vert x-y\Vert _2\) for all \(x,y \in {\mathbb {R}}^n\). A famous concentration inequality by Sudakov, Tsirelson, Borell states that \(\Pr [|F(g)-\mu | \ge t] \le e^{-t^2/\pi ^2}\), where g is a random Gaussian and \(\mu \) is the mean of F under g.
References
Agmon, S.: The relaxation method for linear inequalities. Can. J. Math. 6, 382–392 (1954)
Arora, S., Hazan, E., Kale, S.: Fast algorithms for approximate semidefinite programming using the multiplicative weights update method. In: 46th IEEE FOCS, pp. 339–348 (2005)
Arora, S., Hazan, E., Kale, S.: The multiplicative weights update method: a meta-algorithm and applications. Theor. Comp. 8, 121–164 (2012)
Alon, N., Spencer, J.H.: The Probabilistic Method. Wiley Series in Discrete Mathematics and Optimization. Wiley, Hoboken (2004)
Ball, K.: An elementary introduction to modern convex geometry. In: Silvio, L. (ed.) Flavors of Geometry, pp. 1–58. University Press, Cambridge (1997)
Betke, U.: Relaxation, new combinatorial and polynomial algorithms for the linear feasibility problem. Discrete Comput. Geom. 32(3), 317–338 (2004)
Conforti, M., Cornuejols, G., Zambelli, G.: Integer Programming. Springer Publishing Company Inc., Heidelberg (2014)
Chubanov, S.: A strongly polynomial algorithm for linear systems having a binary solution. Math. Program. 134(2), 533–570 (2012)
Chubanov, S.: A polynomial projection algorithm for linear feasibility problems. Math. Program. 153(2), 687–713 (2015)
Christiano, P., Kelner, J.A., Madry, A., Spielman, D.A., Teng, S.: Electrical flows, laplacian systems, and faster approximation of maximum flow in undirected graphs. In: Proceedings of the 43rd ACM Symposium on Theory of Computing, New York, NY, USA, pp. 273–282 (2011)
Dantzig, G.B.: Maximization of a linear function of variables subject to linear inequalities. In: Activity Analysis of Production and Allocation, Cowles Commission Monograph, vol. 13, pp. 339–347. John Wiley & Sons Inc., Chapman & Hall Ltd., New York (1951)
Dunagan, J., Vempala, S.: A simple polynomial-time rescaling algorithm for solving linear programs. Math. Program. 114(1), 101–114 (2006)
Dadush, D., Végh, L.A., Zambelli, G.: Rescaling algorithms for linear programming - part I: conic feasibility. CoRR, abs/1611.06427 (2016)
Garg, N., Könemann, J.: Faster and simpler algorithms for multicommodity flow and other fractional packing problems. SIAM J. Comput. 37(2), 630–652 (2007)
Hačijan, L.G.: A polynomial algorithm in linear programming. Dokl. Akad. Nauk SSSR 244(5), 1093–1096 (1979)
John, F.: Extremum problems with inequalities as subsidiary conditions. In: Friedrichs, K.O., Neugebauer, O.E., Stoker, J.J. (eds.) Studies and Essays presented to R. Courant on his 60th Birthday, pp. 187–204. Interscience Publishers, New York (1948)
Karmarkar, N.: A new polynomial-time algorithm for linear programming. Combinatorica 4(4), 373–395 (1984)
Klee, V., Minty, G.: How good is the simplex algorithm? In: Inequalities, III (Proceedings Third Symposium, UCLA, 1969; Dedicated to the Memory of Theodore S. Motzkin), pp. 159–175. Academic Press, New York (1972)
Lee, Y., Sinford, A.: A new polynomial-time algorithm for linear programming (2015). https://arxiv.org/abs/1312.6677
Madry, A.: Faster approximation schemes for fractional multicommodity flow problems via dynamic graph algorithms. In: Proceedings of the 42nd ACM Symposium on Theory of Computing, New York, NY, pp. 121–130 (2010)
Nesterov, Y.: Excessive gap technique in nonsmooth convex minimization. SIAM J. Optim. 16(1), 235–249 (2005)
Peña, J., Soheili, N.: A smooth perceptron algorithm. SIAM J. Optim. 22(2), 728–737 (2012)
Peña, J., Soheili, N.: A deterministic rescaled perceptron algorithm. Math. Program. 155(1–2), 497–510 (2016)
Plotkin, S.A., Shmoys, D.B., Tardos, E.: Fast approximation algorithms for fractional packing and covering problems. Math. Oper. Res. 20(2), 257–301 (1995)
Schrijver, A.: Theory of linear and integer programming. Wiley-Interscience Series in Discrete Mathematics. John Wiley and Sons, Inc., New York (1986)
Vazirani, V.: Approximation Algorithms. Springer, Heidelberg (2001)
Williamson, D.P., Shmoys, D.B.: The Design of Approximation Algorithms. University Press, Cambridge (2011)
Author information
Authors and Affiliations
Corresponding author
Editor information
Editors and Affiliations
Rights and permissions
Copyright information
© 2017 Springer International Publishing AG
About this paper
Cite this paper
Hoberg, R., Rothvoss, T. (2017). An Improved Deterministic Rescaling for Linear Programming Algorithms. In: Eisenbrand, F., Koenemann, J. (eds) Integer Programming and Combinatorial Optimization. IPCO 2017. Lecture Notes in Computer Science(), vol 10328. Springer, Cham. https://doi.org/10.1007/978-3-319-59250-3_22
Download citation
DOI: https://doi.org/10.1007/978-3-319-59250-3_22
Published:
Publisher Name: Springer, Cham
Print ISBN: 978-3-319-59249-7
Online ISBN: 978-3-319-59250-3
eBook Packages: Computer ScienceComputer Science (R0)