Abstract
Combining machine learning and constrained optimization, Predict+ Optimize tackles optimization problems containing parameters that are unknown at the time of solving. Prior works focus on cases with unknowns only in the objectives. A new framework was recently proposed to cater for unknowns also in constraints by introducing a loss function, called Post-hoc Regret, that takes into account the cost of correcting an unsatisfiable prediction. Since Post-hoc Regret is non-differentiable, the previous work computes only its approximation. While the notion of Post-hoc Regret is general, its specific implementation is applicable to only packing and covering linear programming problems. In this paper, we first show how to compute Post-hoc Regret exactly for any optimization problem solvable by a recursive algorithm satisfying simple conditions. Experimentation demonstrates substantial improvement in the quality of solutions as compared to the earlier approximation approach. Furthermore, we show experimentally the empirical behavior of different combinations of correction and penalty functions used in the Post-hoc Regret of the same benchmarks. Results provide insights for defining the appropriate Post-hoc Regret in different application scenarios.
Access this chapter
Tax calculation will be finalised at checkout
Purchases are for personal use only
References
Cappart, Q., Moisan, T., Rousseau, L.M., Prémont-Schwarz, I., Cire, A.A.: Combining reinforcement learning and constraint programming for combinatorial optimization. In: Proceedings of the Thirty-Fourth AAAI Conference on Artificial Intelligence, vol. 35, pp. 3677–3687 (2021)
Collet, M., Gotlieb, A., Lazaar, N., Carlsson, M., Marijan, D., Mossige, M.: RobTest: a CP approach to generate maximal test trajectories for industrial robots. In: Simonis, H. (ed.) CP 2020. LNCS, vol. 12333, pp. 707–723. Springer, Cham (2020). https://doi.org/10.1007/978-3-030-58475-7_41
Demirović, E., et al.: Predict+Optimise with ranking objectives: exhaustively learning linear functions. In: Proceedings of the Twenty-Eighth International Joint Conference on Artificial Intelligence, pp. 1078–1085 (2019)
Demirović, E., et al.: Dynamic programming for Predict+Optimise. In: Proceedings of the Thirty-Fourth AAAI Conference on Artificial Intelligence, pp. 1444–1451 (2020)
Elmachtoub, A.N., Grigas, P.: Smart predict, then optimize. Manag. Sci. 68(1), 9–26 (2022)
Friedman, J., Hastie, T., Tibshirani, R.: The Elements of Statistical Learning, vol. 1, Number 10. Springer series in statistics, New York (2001). https://doi.org/10.1007/978-0-387-21606-5
Genc, B., O’Sullivan, B.: A two-phase constraint programming model for examination timetabling at university college cork. In: Simonis, H. (ed.) CP 2020. LNCS, vol. 12333, pp. 724–742. Springer, Cham (2020). https://doi.org/10.1007/978-3-030-58475-7_42
de Givry, S., Lee, J.H.M., Leung, K.L., Shum, Y.W.: Solving a judge assignment problem using conjunctions of global cost functions. In: O’Sullivan, B. (ed.) CP 2014. LNCS, vol. 8656, pp. 797–812. Springer, Cham (2014). https://doi.org/10.1007/978-3-319-10428-7_57
Hu, X., Lee, J.C., Lee, J.H.: Predict+Optimize for packing and covering LPs with unknown parameters in constraints. In: Proceedings of the Thirty-Second AAAI Conference on Artificial Intelligence (2022)
Hu, X., Lee, J.C., Lee, J.H., Zhong, A.Z.: Branch & learn for recursively and iteratively solvable problems in Predict+Optimize. Adv. Neural Inf. Process. Syst. 35 (2022)
Li, D., et al.: A novel method to solve neural knapsack problems. In: Proceedings of the Thirty-Eighth International Conference on Machine Learning, pp. 6414–6424. PMLR (2021)
LLC, M.: Geant topology map dec2018 copy (2018). https://www.geant.org/Resources/Documents/GEANT_Topology_Map_December_2018.pdf. Accessed 10 Sep 2020
Lucerna, D., Gatti, N., Maier, G., Pattavina, A.: On the efficiency of a game theoretic approach to sparse regenerator placement in WDM networks. In: GLOBECOM 2009–2009 IEEE Global Telecommunications Conference, pp. 1–6. IEEE (2009)
Mandi, J., Stuckey, P.J., Guns, T., et al.: Smart predict-and-optimize for hard combinatorial optimization problems. In: Proceedings of the Thirty-Fourth AAAI Conference on Artificial Intelligence, vol. 34, pp. 1603–1610 (2020)
Orlowski, S., Pióro, M., Tomaszewski, A., Wessäly, R.: SNDlib 1.0-Survivable network design library. In: Proceedings of the Third International Network Optimization Conference, April 2007. http://www.zib.de/orlowski/Paper/OrlowskiPioroTomaszewskiWessaely2007-SNDlib-INOC.pdf.gz, http://sndlib.zib.de, extended version accepted in Networks, 2009
Pisinger, D.: Where are the hard knapsack problems? Comput. Oper. Res. 32(9), 2271–2284 (2005)
Simonis, H., O’Sullivan, B., Mehta, D., Hurley, B., Cauwer, M.D.: Energy-Cost Aware Scheduling/Forecasting Competition (2014). http://challenge.icon-fet.eu/sites/default/files/iconchallenge.pdf
Von Mohrenschildt, M.: A normal form for function rings of piecewise functions. J. Symb. Comput. 26(5), 607–619 (1998)
Waissi, G.R.: Network flows: theory, algorithms, and applications (1994)
Wilder, B., Dilkina, B., Tambe, M.: Melding the data-decisions pipeline: decision-focused learning for combinatorial optimization. In: Proceedings of the Thirty-Third AAAI Conference on Artificial Intelligence, pp. 1658–1665 (2019)
Acknowledgments
We thank the anonymous referees for their constructive comments. In addition, Xinyi Hu and Jimmy H.M. Lee acknowledge the financial support of a General Research Fund (RGC Ref. No. CUHK 14206321) by the University Grants Committee, Hong Kong. Jasper C.H. Lee was supported in part by the generous funding of a Croucher Fellowship for Postdoctoral Research, NSF award DMS-2023239, NSF Medium Award CCF-2107079 and NSF AiTF Award CCF-2006206.
Author information
Authors and Affiliations
Corresponding authors
Editor information
Editors and Affiliations
Rights and permissions
Copyright information
© 2023 The Author(s), under exclusive license to Springer Nature Switzerland AG
About this paper
Cite this paper
Hu, X., Lee, J.C.H., Lee, J.H.M. (2023). Branch & Learn with Post-hoc Correction for Predict+Optimize with Unknown Parameters in Constraints. In: Cire, A.A. (eds) Integration of Constraint Programming, Artificial Intelligence, and Operations Research. CPAIOR 2023. Lecture Notes in Computer Science, vol 13884. Springer, Cham. https://doi.org/10.1007/978-3-031-33271-5_18
Download citation
DOI: https://doi.org/10.1007/978-3-031-33271-5_18
Published:
Publisher Name: Springer, Cham
Print ISBN: 978-3-031-33270-8
Online ISBN: 978-3-031-33271-5
eBook Packages: Computer ScienceComputer Science (R0)