Papers
Topics
Authors
Recent
Gemini 2.5 Flash
Gemini 2.5 Flash
110 tokens/sec
GPT-4o
56 tokens/sec
Gemini 2.5 Pro Pro
44 tokens/sec
o3 Pro
6 tokens/sec
GPT-4.1 Pro
47 tokens/sec
DeepSeek R1 via Azure Pro
28 tokens/sec
2000 character limit reached

Machine Learning Insides OptVerse AI Solver: Design Principles and Applications (2401.05960v2)

Published 11 Jan 2024 in cs.AI

Abstract: In an era of digital ubiquity, efficient resource management and decision-making are paramount across numerous industries. To this end, we present a comprehensive study on the integration of ML techniques into Huawei Cloud's OptVerse AI Solver, which aims to mitigate the scarcity of real-world mathematical programming instances, and to surpass the capabilities of traditional optimization techniques. We showcase our methods for generating complex SAT and MILP instances utilizing generative models that mirror multifaceted structures of real-world problem. Furthermore, we introduce a training framework leveraging augmentation policies to maintain solvers' utility in dynamic environments. Besides the data generation and augmentation, our proposed approaches also include novel ML-driven policies for personalized solver strategies, with an emphasis on applications like graph convolutional networks for initial basis selection and reinforcement learning for advanced presolving and cut selection. Additionally, we detail the incorporation of state-of-the-art parameter tuning algorithms which markedly elevate solver performance. Compared with traditional solvers such as Cplex and SCIP, our ML-augmented OptVerse AI Solver demonstrates superior speed and precision across both established benchmarks and real-world scenarios, reinforcing the practical imperative and effectiveness of machine learning techniques in mathematical programming solvers.

Definition Search Book Streamline Icon: https://streamlinehq.com
References (118)
  1. Huawei Cloud. The optverse ai solver empowers the tianjin port intelligent planning platform, realizing global integrated planning, https://www.huaweicloud.com/cases/tjg.html, 2023.
  2. ZIB. Scip solver, https://www.scipopt.org/, 2021.
  3. Bob Bixby. The gurobi optimizer. Transp. Re-search Part B, 41(2):159–178, 2007.
  4. Solving mixed-integer quadratic programming problems with ibm-cplex: a progress report. In Proceedings of the twenty-sixth RAMP symposium, pages 16–17, 2014.
  5. Hebo: Pushing the limits of sample-efficient hyper-parameter optimisation. Journal of Artificial Intelligence Research, 74(1):1269–1349, 2022.
  6. End-to-end meta-bayesian optimisation with transformer neural processes. https://arxiv.org/abs/2305.15930, 2023.
  7. Pointer networks. Advances in Neural Information Processing Systems, 28, 2015.
  8. Neural combinatorial optimization with reinforcement learning. arXiv preprint arXiv:1611.09940, 2016.
  9. Attention, learn to solve routing problems! In International Conference on Learning Representations, 2019. URL https://openreview.net/forum?id=ByxBFsRqYm.
  10. Learning a SAT solver from single-bit supervision. In International Conference on Learning Representations, 2019. URL https://openreview.net/forum?id=HJMC_iA5tm.
  11. Learning local search heuristics for boolean satisfiability. Advances in Neural Information Processing Systems, 32, 2019.
  12. Reinforcement learning for solving the vehicle routing problem. Advances in Neural Information Processing Systems, 31, 2018.
  13. Exploratory combinatorial optimization with reinforcement learning. In Proceedings of the AAAI Conference on Artificial Intelligence, volume 34, pages 3243–3250, 2020.
  14. T2t: From distribution learning in training to gradient search in testing for combinatorial optimization. In Advances in Neural Information Processing Systems, 2023a.
  15. Accelerating linear programming solving by exploiting the performance variability via reinforcement learning. 2023b.
  16. Accelerate presolve in large-scale linear programming via reinforcement learning. arXiv preprint arXiv:2310.11845, 2023a.
  17. Machine learning for combinatorial optimization: a methodological tour d’horizon. European Journal of Operational Research, 290(2):405–421, 2021.
  18. Machine learning for combinatorial optimization, 2021. URL https://www.ecole.ai/2021/ml4co-competition/.
  19. Tobias Achterberg. Constraint integer programming. 2007.
  20. Reinforcement learning for integer programming: Learning to cut. In Hal Daumé III and Aarti Singh, editors, Proceedings of the 37th International Conference on Machine Learning, volume 119 of Proceedings of Machine Learning Research, pages 9367–9376. PMLR, 13–18 Jul 2020.
  21. Learning to cut by looking ahead: Cutting plane selection via imitation learning. In Kamalika Chaudhuri, Stefanie Jegelka, Le Song, Csaba Szepesvari, Gang Niu, and Sivan Sabato, editors, Proceedings of the 39th International Conference on Machine Learning, volume 162 of Proceedings of Machine Learning Research, pages 17584–17600. PMLR, 17–23 Jul 2022.
  22. Adaptive cut selection in mixed-integer linear programming. arXiv preprint arXiv:2202.10962, 2022.
  23. Scoring positive semidefinite cutting planes for quadratic optimization via trained neural networks. preprint: http://www. optimization-online. org/DB_ HTML/2018/11/6943. html, 2019.
  24. Learning to branch in mixed integer programming. Proceedings of the AAAI Conference on Artificial Intelligence, 30(1), Feb. 2016. doi: 10.1609/aaai.v30i1.10080. URL https://ojs.aaai.org/index.php/AAAI/article/view/10080.
  25. Learning to branch. In Jennifer Dy and Andreas Krause, editors, Proceedings of the 35th International Conference on Machine Learning, volume 80 of Proceedings of Machine Learning Research, pages 344–353. PMLR, 10–15 Jul 2018.
  26. Parameterizing branch-and-bound search trees to learn branching policies. In Proceedings of the AAAI Conference on Artificial Intelligence, volume 35, pages 3931–3939, 2021.
  27. Learning to search in branch and bound algorithms. In Z. Ghahramani, M. Welling, C. Cortes, N. Lawrence, and K.Q. Weinberger, editors, Advances in Neural Information Processing Systems, volume 27. Curran Associates, Inc., 2014.
  28. Guiding combinatorial optimization with uct. In International conference on integration of artificial intelligence (AI) and operations research (OR) techniques in constraint programming, pages 356–361. Springer, 2012.
  29. Machine-learning-based column selection for column generation. Transportation Science, 55(4):815–831, 2021.
  30. Learning to run heuristics in tree search. In Ijcai, pages 659–666, 2017.
  31. Adaptive algorithmic behavior for solving mixed integer programs using bandit algorithms. In Operations Research Proceedings 2018, pages 513–519. Springer, 2019.
  32. Accelerating primal solution findings for mixed integer programs based on solution prediction. In AAAI, volume 34, pages 1452–1459, 2020. doi: 10.1609/AAAI.V34I02.5503.
  33. Hybrid models for learning to branch. NeurIPS, 33:18087–18097, 2020.
  34. Lookback for learning to branch. arXiv:2206.14987, 2022.
  35. An improved reinforcement learning algorithm for learning to branch. arXiv:2201.06213, 2022.
  36. Learning to reformulate for linear programming. arXiv:2201.06216, 2022.
  37. On representing linear programs by graph neural networks. arXiv:2209.12288, 2022.
  38. A systematic survey on deep generative models for graph generation. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2022.
  39. Graph networks for molecular design. Machine Learning: Science and Technology, 2(2):025023, 2021.
  40. Variational graph auto-encoders. arXiv preprint arXiv:1611.07308, 2016.
  41. Generative diffusion models on graphs: Methods and applications. arXiv preprint arXiv:2302.02591, 2023a.
  42. A survey on deep graph generation: Methods and applications. arXiv preprint arXiv:2203.06714, 2022.
  43. Masked graph modeling for molecule generation. Nature communications, 12(1):3156, 2021.
  44. Junction tree variational autoencoder for molecular graph generation. In International conference on machine learning, pages 2323–2332. PMLR, 2018.
  45. De novo molecular generation via connection-aware motif mining. In The Eleventh International Conference on Learning Representations, 2023a.
  46. Collective dynamics of ‘small-world’networks. nature, 393(6684):440–442, 1998.
  47. Kronecker graphs: an approach to modeling networks. Journal of Machine Learning Research, 11(2), 2010.
  48. G2sat: learning to generate sat formulas. Advances in neural information processing systems, 32, 2019.
  49. Hardsatgen: Understanding the difficulty of hard sat formula generation and a strong structure-hardness-aware baseline. In Proceedings of the 29th ACM SIGKDD Conference on Knowledge Discovery and Data Mining, 2023c.
  50. On the performance of deep generative models of realistic sat instances. In 25th International Conference on Theory and Applications of Satisfiability Testing (SAT 2022). Schloss Dagstuhl-Leibniz-Zentrum für Informatik, 2022.
  51. Heuristic bounds and test problem generation for the time-dependent traveling salesman problem. Transportation Science, 29(2):167–183, 1995.
  52. Set covering algorithms using cutting planes, heuristics, and subgradient optimization: A computational study, pages 37–60. Springer Berlin Heidelberg, Berlin, Heidelberg, 1980. ISBN 978-3-642-00802-3. doi: 10.1007/BFb0120886. URL https://doi.org/10.1007/BFb0120886.
  53. Alper Atamtürk. On the facets of the mixed–integer knapsack polyhedron. Mathematical Programming, 98(1-3):145–175, 2003.
  54. Simon Andrew Bowly. Stress testing mixed integer programming solvers through new test instance generation methods. PhD thesis, School of Mathematical Sciences, Monash University, 2019.
  55. Generalization of neural combinatorial solvers through the lens of adversarial robustness. In International Conference on Learning Representations, 2022. URL https://openreview.net/forum?id=vJZ7dPIjip3.
  56. Roco: A general framework for evaluating robustness of combinatorial optimization solvers on graphs. In The Eleventh International Conference on Learning Representations, 2023.
  57. Asp: Learn a universal neural solver! arXiv preprint arXiv:2303.00466, 2023a.
  58. Exact combinatorial optimization with graph convolutional neural networks. Advances in neural information processing systems, 32, 2019a.
  59. Theoretical challenges towards cutting-plane selection. Mathematical Programming, 170(1):237–266, 2018.
  60. Franz Wesselmann and U Stuhl. Implementing cutting plane management and selection techniques. In Technical Report. University of Paderborn, 2012.
  61. Learning to select cuts for efficient mixed-integer programming. Pattern Recognition, 123:108353, 2022. ISSN 0031-3203. doi: https://doi.org/10.1016/j.patcog.2021.108353.
  62. Ralph Gomory. An algorithm for the mixed integer problem. Technical report, RAND CORP SANTA MONICA CA, 1960.
  63. Sample complexity of tree search configuration: Cutting planes and beyond. Advances in Neural Information Processing Systems, 34:4015–4027, 2021.
  64. Structural analysis of branch-and-cut and the learnability of gomory mixed integer cuts. arXiv preprint arXiv:2204.07312, 2022.
  65. Practical bayesian optimization of machine learning algorithms. In NIPS, pages 2951–2959, 2012.
  66. Hyperopt: a python library for model selection and hyperparameter optimization. Computational Science & Discovery, 8(1):14008, 2015.
  67. Algorithms for hyper-parameter optimization. In NIPS, pages 2546–2554, 2011.
  68. Sequential model-based optimization for general algorithm configuration. In LION, pages 507–523, 2011.
  69. Google vizier: A service for black-box optimization. In Proceedings of the 23rd ACM SIGKDD International Conference on Knowledge Discovery and Data Mining, page 1487–1495, 2017.
  70. Parallelizing exploration-exploitation tradeoffs in gaussian process bandit optimization. Journal of Machine Learning Research, 15(1):3873–3923, 2014.
  71. Optuna: A next-generation hyperparameter optimization framework. In Proceedings of the 25th ACM SIGKDD International Conference on Knowledge Discovery and Data Mining, page 2623–2631, 2019.
  72. Microsoft. Microsoft nni, https://https://github.com/microsoft/nni, 2021.
  73. Gurobi. Gurobi tuning tool, https://www.gurobi.com/documentation/current/refman/parameter_tuning_tool.html, 2018.
  74. IBM. Cplex tuning tool, https://www.ibm.com/docs/en/icos/20.1.0?topic=programmingconsiderations-tuning-tool, 2021.
  75. Mindopt tuner: Boost the performance of numerical software by automatic parameter tuning. https://arxiv.org/abs/2307.08085, 2023.
  76. A deep instance generative framework for milp solvers under limited data availability. In Advances in Neural Information Processing Systems, 2023b.
  77. G2MILP: Learning to Generate Mixed-Integer Linear Programming Instances for MILP Solvers. 11 2023b. doi: 10.36227/techrxiv.24566554.v1. URL https://www.techrxiv.org/articles/preprint/G2MILP_Learning_to_Generate_Mixed-Integer_Linear_Programming_Instances_for_MILP_Solvers/24566554.
  78. Proximal policy optimization algorithms. arXiv preprint arXiv:1707.06347, 2017.
  79. Promoting generalization for exact solvers via adversarial instance augmentation. arXiv preprint arXiv:2310.14161, 2023.
  80. Decomposition principle for linear programs. Oper. Res., 8(1):101–111, 1960.
  81. Robert E Bixby. Implementing the simplex method: The initial basis. INFORMS J. Comput., 4(3):267–284, 1992.
  82. New crash procedures for large systems of linear constraints. Math. Program., 45(1):475–501, 1989.
  83. Hélcio Vieira Junior and Marcos Pereira Estellita Lins. An improved initial basis for the simplex algorithm. Comput. Oper. Res., 32(8):1983–1993, 2005.
  84. A triangulation and fill-reducing initialization procedure for the simplex algorithm. Math. Program. Comput., 13:491–508, 2021.
  85. The ‘Idiot’ crash quadratic penalty algorithm for linear programming and its application to linearizations of quadratic assignment problems. Optim. Methods Software, 35(3):488–501, 2020. doi: 10.1080/10556788.2019.1604702.
  86. Weisfeiler and Leman go neural: Higher-order graph neural networks. In AAAI, volume 33, pages 4602–4609, 2019. doi: 10.1609/aaai.v33i01.33014602.
  87. Knowledge-injected federated learning. arXiv:2208.07530, 2022.
  88. Smart initial basis selection for linear programs. In Proceedings of the 40th International Conference on Machine Learning, volume 202 of Proceedings of Machine Learning Research, pages 9650–9664. PMLR, 23–29 Jul 2023b.
  89. MIRPLib–a library of maritime inventory routing problem instances: Survey, core model, and benchmark results. Eur. J. Oper. Res., 235(2):350–366, 2014.
  90. 1−limit-from11-1 -norm support vector machines. NeurIPS, 16, 2003.
  91. Practical large-scale linear programming using primal-dual hybrid gradient. NeurIPS, 34:20243–20257, 2021.
  92. Jordi Castro and Paula de la Lama-Zubirán. A new interior-point approach for large separable convex quadratic two-stage stochastic problems. Optim. Methods Software, pages 1–29, 2020.
  93. Generation techniques for linear programming instances with controllable properties. Math. Program. Comput., 12(3):389–415, 2020. doi: 10.1007/s12532-019-00170-6.
  94. István Maros. Computational techniques of the simplex method, volume 61. Springer Science & Business Media, 2002.
  95. Joseph M Elble. Computational experience with linear optimization and related problems. University of Illinois at Urbana-Champaign, 2010.
  96. Advanced preprocessing techniques for linear and quadratic programming. OR Spectr., 25(4):575–595, 2003. doi: 10.1007/s00291-003-0130-x. URL https://doi.org/10.1007/s00291-003-0130-x.
  97. Presolving in linear programming. Mathematical Programming, 71(2):221–245, 1995.
  98. coin-or/clp: Release releases/1.17.7, January 2022.
  99. Ivet Galabova. Presolve, crash and software engineering for highs. 2023.
  100. Accelerate presolve in large-scale linear programming via reinforcement learning. arXiv preprint arXiv:2310.11845, 2023b.
  101. Learning cut selection for mixed-integer linear programming via hierarchical sequence model. In The Eleventh International Conference on Learning Representations, 2023c. URL https://openreview.net/forum?id=Zob4P9bRNcK.
  102. Asynchronous methods for deep reinforcement learning. In Maria Florina Balcan and Kilian Q. Weinberger, editors, Proceedings of The 33rd International Conference on Machine Learning, volume 48 of Proceedings of Machine Learning Research, pages 1928–1937, New York, New York, USA, 20–22 Jun 2016. PMLR.
  103. Learning cut selection for mixed-integer linear programming via hierarchical sequence model. In International Conference on Learning Representations, 2023d. URL https://openreview.net/forum?id=Zob4P9bRNcK.
  104. Solving mixed integer programs using neural networks. arXiv preprint arXiv:2012.13349, 2020.
  105. Exact combinatorial optimization with graph convolutional neural networks. Advances in neural information processing systems, 32, 2019b.
  106. Timo Berthold. Measuring the impact of primal heuristics. Operations Research Letters, 41(6):611–614, 2013.
  107. Journal of Global Optimization, 11(4):341–359, 1997. doi: 10.1023/a:1008202821328. URL https://doi.org/10.1023/a:1008202821328.
  108. Warped gaussian processes. In NIPS, 2003.
  109. Transfer learning for bayesian optimization: A survey. https://arxiv.org/abs/2302.05927, 2023.
  110. Recent advances in differential evolution – an updated survey. Swarm and Evolutionary Computation, 27:1–30, April 2016. doi: 10.1016/j.swevo.2016.01.004. URL https://doi.org/10.1016/j.swevo.2016.01.004.
  111. Exploration and exploitation in evolutionary algorithms. ACM Computing Surveys, 45(3):1–33, June 2013. doi: 10.1145/2480741.2480752. URL https://doi.org/10.1145/2480741.2480752.
  112. JADE: Adaptive differential evolution with optional external archive. IEEE Transactions on Evolutionary Computation, 13(5):945–958, October 2009a. doi: 10.1109/tevc.2009.2014613. URL https://doi.org/10.1109/tevc.2009.2014613.
  113. Success-history based parameter adaptation for differential evolution. In 2013 IEEE Congress on Evolutionary Computation, pages 71–78, 2013. doi: 10.1109/CEC.2013.6557555.
  114. Improving the search performance of shade using linear population size reduction. In 2014 IEEE Congress on Evolutionary Computation (CEC), pages 1658–1665, 2014. doi: 10.1109/CEC.2014.6900380.
  115. Adaptive Differential Evolution. Springer Berlin Heidelberg, 2009b. doi: 10.1007/978-3-642-01527-4. URL https://doi.org/10.1007/978-3-642-01527-4.
  116. JADE: Self-adaptive differential evolution with fast and reliable convergence performance. In 2007 IEEE Congress on Evolutionary Computation. IEEE, September 2007. doi: 10.1109/cec.2007.4424751. URL https://doi.org/10.1109/cec.2007.4424751.
  117. Multi-start JADE with knowledge transfer for numerical optimization. In 2009 IEEE Congress on Evolutionary Computation. IEEE, May 2009. doi: 10.1109/cec.2009.4983171. URL https://doi.org/10.1109/cec.2009.4983171.
  118. J. H. Halton. Algorithm 247: Radical-inverse quasi-random point sequence. Communications of the ACM, 7(12):701–702, December 1964. doi: 10.1145/355588.365104. URL https://doi.org/10.1145/355588.365104.
User Edit Pencil Streamline Icon: https://streamlinehq.com
Authors (26)
  1. Xijun Li (25 papers)
  2. Fangzhou Zhu (21 papers)
  3. Hui-Ling Zhen (33 papers)
  4. Weilin Luo (9 papers)
  5. Meng Lu (28 papers)
  6. Yimin Huang (17 papers)
  7. Zhenan Fan (16 papers)
  8. Zirui Zhou (32 papers)
  9. Yufei Kuang (10 papers)
  10. Zhihai Wang (52 papers)
  11. Zijie Geng (9 papers)
  12. Yang Li (1142 papers)
  13. Haoyang Liu (45 papers)
  14. Zhiwu An (1 paper)
  15. Muming Yang (2 papers)
  16. Jianshu Li (34 papers)
  17. Jie Wang (480 papers)
  18. Junchi Yan (241 papers)
  19. Defeng Sun (81 papers)
  20. Tao Zhong (66 papers)
Citations (1)
View on Semantic Scholar

Summary

We haven't generated a summary for this paper yet.

Ai Sparkles Streamline Icon: https://streamlinehq.com Summarize Now
X Twitter Logo Streamline Icon: https://streamlinehq.com

Tweets