IDEAS home Printed from https://ideas.repec.org/a/spr/jglopt/v85y2023i3d10.1007_s10898-022-01228-x.html
   My bibliography  Save this article

Convex and concave envelopes of artificial neural network activation functions for deterministic global optimization

Author

Listed:
  • Matthew E. Wilhelm

    (University of Connecticut)

  • Chenyu Wang

    (University of Connecticut)

  • Matthew D. Stuber

    (University of Connecticut)

Abstract

In this work, we present general methods to construct convex/concave relaxations of the activation functions that are commonly chosen for artificial neural networks (ANNs). The choice of these functions is often informed by both broader modeling considerations balanced with a need for high computational performance. The direct application of factorable programming techniques to compute bounds and convex/concave relaxations of such functions often lead to weak enclosures due to the dependency problem. Moreover, the piecewise formulation that defines several popular activation functions, prevents the computation of convex/concave relaxations as they violate the factorable function requirement. To improve the performance of relaxations of ANNs for deterministic global optimization applications, this study presents the development of a library of envelopes of the thoroughly studied rectifier-type and sigmoid activation functions, in addition to the novel self-gated sigmoid-weighted linear unit (SiLU) and Gaussian error linear unit activation functions. We demonstrate that the envelopes of activation functions directly lead to tighter relaxations of ANNs on their input domain. In turn, these improvements translate to a dramatic reduction in CPU runtime required for solving optimization problems involving ANN models to epsilon-global optimality. We further demonstrate that the factorable programming approach leads to superior computational performance over alternative state-of-the-art approaches.

Suggested Citation

  • Matthew E. Wilhelm & Chenyu Wang & Matthew D. Stuber, 2023. "Convex and concave envelopes of artificial neural network activation functions for deterministic global optimization," Journal of Global Optimization, Springer, vol. 85(3), pages 569-594, March.
  • Handle: RePEc:spr:jglopt:v:85:y:2023:i:3:d:10.1007_s10898-022-01228-x
    DOI: 10.1007/s10898-022-01228-x
    as

    Download full text from publisher

    File URL: http://link.springer.com/10.1007/s10898-022-01228-x
    File Function: Abstract
    Download Restriction: Access to the full text of the articles in this series is restricted.

    File URL: https://libkey.io/10.1007/s10898-022-01228-x?utm_source=ideas
    LibKey link: if access is restricted and if your library uses this service, LibKey will redirect you to where you can use your library subscription to access this item
    ---><---

    As the access to this document is restricted, you may want to search for a different version of it.

    References listed on IDEAS

    as
    1. Rohit Kannan & Paul I. Barton, 2017. "The cluster problem in constrained global optimization," Journal of Global Optimization, Springer, vol. 69(3), pages 629-676, November.
    2. Jaromił Najman & Alexander Mitsos, 2016. "Convergence analysis of multivariate McCormick relaxations," Journal of Global Optimization, Springer, vol. 66(4), pages 597-628, December.
    3. Yongho Seong & Changhyup Park & Jinho Choi & Ilsik Jang, 2020. "Surrogate Model with a Deep Neural Network to Evaluate Gas–Liquid Flow in a Horizontal Pipe," Energies, MDPI, vol. 13(4), pages 1-12, February.
    4. Achim Wechsung & Spencer Schaber & Paul Barton, 2014. "The cluster problem revisited," Journal of Global Optimization, Springer, vol. 58(3), pages 429-438, March.
    5. Dominik Bongartz & Alexander Mitsos, 2017. "Deterministic global optimization of process flowsheets in a reduced space using McCormick relaxations," Journal of Global Optimization, Springer, vol. 69(4), pages 761-796, December.
    6. Celik, Ali N. & Kolhe, Mohan, 2013. "Generalized feed-forward based method for wind energy prediction," Applied Energy, Elsevier, vol. 101(C), pages 582-588.
    7. Ralph Kearfott & Jessie Castille & Gaurav Tyagi, 2013. "A general framework for convexity analysis in deterministic global optimization," Journal of Global Optimization, Springer, vol. 56(3), pages 765-785, July.
    8. Kamil A. Khan & Matthew Wilhelm & Matthew D. Stuber & Huiyi Cao & Harry A. J. Watson & Paul I. Barton, 2018. "Corrections to: Differentiable McCormick relaxations," Journal of Global Optimization, Springer, vol. 70(3), pages 705-706, March.
    9. Joseph Scott & Paul Barton, 2013. "Improved relaxations for the parametric solutions of ODEs using differential inequalities," Journal of Global Optimization, Springer, vol. 57(1), pages 143-176, September.
    10. Joseph Scott & Matthew Stuber & Paul Barton, 2011. "Generalized McCormick relaxations," Journal of Global Optimization, Springer, vol. 51(4), pages 569-606, December.
    Full references (including those not matched with items on IDEAS)

    Most related items

    These are the items that most often cite the same works as this one and are cited by the same works as this one.
    1. Matthew E. Wilhelm & Matthew D. Stuber, 2023. "Improved Convex and Concave Relaxations of Composite Bilinear Forms," Journal of Optimization Theory and Applications, Springer, vol. 197(1), pages 174-204, April.
    2. Artur M. Schweidtmann & Alexander Mitsos, 2019. "Deterministic Global Optimization with Artificial Neural Networks Embedded," Journal of Optimization Theory and Applications, Springer, vol. 180(3), pages 925-948, March.
    3. Jaromił Najman & Alexander Mitsos, 2019. "On tightness and anchoring of McCormick and other relaxations," Journal of Global Optimization, Springer, vol. 74(4), pages 677-703, August.
    4. Spencer D. Schaber & Joseph K. Scott & Paul I. Barton, 2019. "Convergence-order analysis for differential-inequalities-based bounds and relaxations of the solutions of ODEs," Journal of Global Optimization, Springer, vol. 73(1), pages 113-151, January.
    5. Kamil A. Khan & Harry A. J. Watson & Paul I. Barton, 2017. "Differentiable McCormick relaxations," Journal of Global Optimization, Springer, vol. 67(4), pages 687-729, April.
    6. Jaromił Najman & Alexander Mitsos, 2019. "Tighter McCormick relaxations through subgradient propagation," Journal of Global Optimization, Springer, vol. 75(3), pages 565-593, November.
    7. Huiyi Cao & Kamil A. Khan, 2023. "General convex relaxations of implicit functions and inverse functions," Journal of Global Optimization, Springer, vol. 86(3), pages 545-572, July.
    8. Chrysoula D. Kappatou & Dominik Bongartz & Jaromił Najman & Susanne Sass & Alexander Mitsos, 2022. "Global dynamic optimization with Hammerstein–Wiener models embedded," Journal of Global Optimization, Springer, vol. 84(2), pages 321-347, October.
    9. Rohit Kannan & Paul I. Barton, 2018. "Convergence-order analysis of branch-and-bound algorithms for constrained problems," Journal of Global Optimization, Springer, vol. 71(4), pages 753-813, August.
    10. N. Kazazakis & C. S. Adjiman, 2018. "Arbitrarily tight $$\alpha $$ α BB underestimators of general non-linear functions over sub-optimal domains," Journal of Global Optimization, Springer, vol. 71(4), pages 815-844, August.
    11. Dominik Bongartz & Alexander Mitsos, 2017. "Deterministic global optimization of process flowsheets in a reduced space using McCormick relaxations," Journal of Global Optimization, Springer, vol. 69(4), pages 761-796, December.
    12. Boukouvala, Fani & Misener, Ruth & Floudas, Christodoulos A., 2016. "Global optimization advances in Mixed-Integer Nonlinear Programming, MINLP, and Constrained Derivative-Free Optimization, CDFO," European Journal of Operational Research, Elsevier, vol. 252(3), pages 701-727.
    13. Jason Ye & Joseph K. Scott, 2023. "Extended McCormick relaxation rules for handling empty arguments representing infeasibility," Journal of Global Optimization, Springer, vol. 87(1), pages 57-95, September.
    14. Rohit Kannan & Paul I. Barton, 2017. "The cluster problem in constrained global optimization," Journal of Global Optimization, Springer, vol. 69(3), pages 629-676, November.
    15. Miten Mistry & Dimitrios Letsios & Gerhard Krennrich & Robert M. Lee & Ruth Misener, 2021. "Mixed-Integer Convex Nonlinear Optimization with Gradient-Boosted Trees Embedded," INFORMS Journal on Computing, INFORMS, vol. 33(3), pages 1103-1119, July.
    16. Jaromił Najman & Alexander Mitsos, 2016. "Convergence analysis of multivariate McCormick relaxations," Journal of Global Optimization, Springer, vol. 66(4), pages 597-628, December.
    17. Marendet, Antoine & Goldsztejn, Alexandre & Chabert, Gilles & Jermann, Christophe, 2020. "A standard branch-and-bound approach for nonlinear semi-infinite problems," European Journal of Operational Research, Elsevier, vol. 282(2), pages 438-452.
    18. A. Skjäl & T. Westerlund & R. Misener & C. A. Floudas, 2012. "A Generalization of the Classical αBB Convex Underestimation via Diagonal and Nondiagonal Quadratic Terms," Journal of Optimization Theory and Applications, Springer, vol. 154(2), pages 462-490, August.
    19. Huster, Wolfgang R. & Schweidtmann, Artur M. & Mitsos, Alexander, 2020. "Globally optimal working fluid mixture composition for geothermal power cycles," Energy, Elsevier, vol. 212(C).
    20. Zhiwei Liao & Bowen Wang & Wenjuan Tao & Ye Liu & Qiyun Hu, 2024. "Research on Decision Optimization and the Risk Measurement of the Power Generation Side Based on Quantile Data-Driven IGDT," Energies, MDPI, vol. 17(7), pages 1-21, March.

    Corrections

    All material on this site has been provided by the respective publishers and authors. You can help correct errors and omissions. When requesting a correction, please mention this item's handle: RePEc:spr:jglopt:v:85:y:2023:i:3:d:10.1007_s10898-022-01228-x. See general information about how to correct material in RePEc.

    If you have authored this item and are not yet registered with RePEc, we encourage you to do it here. This allows to link your profile to this item. It also allows you to accept potential citations to this item that we are uncertain about.

    If CitEc recognized a bibliographic reference but did not link an item in RePEc to it, you can help with this form .

    If you know of missing items citing this one, you can help us creating those links by adding the relevant references in the same way as above, for each refering item. If you are a registered author of this item, you may also want to check the "citations" tab in your RePEc Author Service profile, as there may be some citations waiting for confirmation.

    For technical questions regarding this item, or to correct its authors, title, abstract, bibliographic or download information, contact: Sonal Shukla or Springer Nature Abstracting and Indexing (email available below). General contact details of provider: http://www.springer.com .

    Please note that corrections may take a couple of weeks to filter through the various RePEc services.

    IDEAS is a RePEc service. RePEc uses bibliographic data supplied by the respective publishers.