Abstract
Bayesian Optimization with an underlying Gaussian Process is used as an optimization solution to a black-box optimization problem in which the function to be optimized has particular properties that result in difficulties. It can only be expressed in terms of a complicated and lengthy stochastic algorithm, with the added complication that the value returned is only required to be sufficiently near to a pre-determined ‘target’. We consider the context of financial stress testing, for which the data used has a significant noise component. Commonly-used Bayesian Optimization acquisition functions cannot analyze the ‘target’ condition in a satisfactory way, but a simple modification of the ‘Lower Confidence Bound’ acquisition function improves results markedly. A proof that the modified acquisition function is superior to the unmodified version is given.
Access this chapter
Tax calculation will be finalised at checkout
Purchases are for personal use only
References
Basel Committee on Banking Supervision, Stress testing principles d450. Bank for International Settlements (BIS) (2018). https://www.bis.org/bcbs/publ/d450.htm
European Banking Authority: 2020 EU-wide stress test methodological note (2019). https://www.eba.europa.eu/sites/default/documents/files/documents/10180/2841396/ba66328f-476f-4707-9a23-6df5957dc8c1/2020%20EU-wide%20stress%20test%20-%20Draft%20Methodological%20Note.pdf
Frachot, A., Georges, P., Roncalli, T.: Loss Distribution Approach for operational risk, Working paper, Groupe de Recherche Operationnelle, Credit Lyonnais, France (2001). https://ssrn.com/abstract=1032523
Mockus, J.: On Bayesian methods for seeking the extremum. In: Proceedings of IFIP Technical Conference, pp. 400–404 (1974). https://dl.acm.org/citation.cfm?id=646296.687872
Mockus, J., Tiesis, V., Zilinskas, A.: The application of Bayesian methods for seeking the extremum. Towards Global Optimisation (eds. Dixon,L. and Szego,G.P.) vol. 2 (1978)
Mockus, J.: The Bayesian approach to local optimization. In: Bayesian Approach to Global Optimization. Mathematics and Its Applications, vol. 37. Springer, Heidelberg (1989). https://doi.org/10.1007/978-94-009-0909-0_7
Cox, D.D., John, S.: SDO: a statistical method for global optimization. In: Multidisciplinary Design Optimization, pp. 315–329. SIAM, Philadelphia (1997)
Srinivas, N., Krause, A., Kakade, S., Seeger, M.: Gaussian process optimization in the bandit setting: no regret and experimental design. In: Proceedings of ICML 2010, pp. 1015–1022 (2010). https://dl.acm.org/citation.cfm?id=3104322.3104451
Rana, S., Li, C., Gupta, S.: High dimensional Bayesian optimization with elastic Gaussian process. In: Proceedings of 34th International Conference on Machine Learning, Sydney, PMLR 70 (2017)
Rasmussen, C.E., Williams, C.K.I.: Gaussian Processes for Machine Learning. MIT Press, Cambridge (2016)
Murphy, K.P.: Machine Learning: A Probabilistic Perspective, Chapter 15. MIT Press, Cambridge (2015)
Berk, J., Nguyen, V., Gupta, S., et al.: Exploration enhanced expected improvement for Bayesian optimization. In: Joint European Conference on Machine Learning and Knowledge Discovery in Databases. LNCS, vol. 11052, pp. 621–637 (2018)
Jones, D.R.: A taxonomy of global optimization methods based on response surfaces. J. Global Optim. 21(4), 345–383 (2001)
Kara, G., Özmen, A., Weber, G.: Stability advances in robust portfolio optimization under parallelepiped uncertainty. Central Eur. J. Oper. Res. 27, 241–261 (2019). https://doi.org/10.1007/s10100-017-0508-5
Özmen, A., Weber, G.W., Batmaz, I., Kropat, E.: RCMARS: robustification of CMARS with different scenarios under polyhedral uncertainty set. Commun. Nonlinear Sci. Numer. Simul. 16(12), 478–4787 (2011). https://doi.org/10.1016/j.cnsns.2011.04.001
Savku, E., Weber, G.: Stochastic differential games for optimal investment problems in a Markov regime-switching jump-diffusion market. Ann. Oper. Res. (2020). https://doi.org/10.1007/s10479-020-03768-5
Kwon, J., Mertikopoulos, P.: A continuous-time approach to online optimization. J. Dyn. Games 4(2), 125–148 (2017). https://doi.org/10.3934/jdg.2017008
Ascher, U.M.: Discrete processes and their continuous limits. J. Dyn. Games 7(2), 123–140 (2020). https://doi.org/10.3934/jdg.2020008
Yang, Y., Sutanto, C.: Chance-constrained optimization for nonconvex programs using scenario-based methods. ISA Trans. 90, 157–168 (2019). https://doi.org/10.1016/j.isatra.2019.01.013
Ozer, F., Toroslu, I.H., Karagoz, P., Yucel, F.: Dynamic programming solution to ATM cash replenishment optimization problem. In: Intelligent Computing & Optimization. ICO 2018, vol. 866. Springer, Cham (2019). https://doi.org/10.1007/978-3-030-00979-3_45
Samakpong, T., Ongsakul, W., Nimal Madhu, M.: Optimal power flow considering cost of wind and solar power uncertainty using particle swarm optimization. In: Intelligent Computing and Optimization. ICO 2019, vol. 1072. Springer, Cham (2020). https://doi.org/10.1007/978-3-030-33585-4_19
Yan, Y.: (2016). https://cran.r-project.org/web/packages/rBayesianOptimization/index.html
Author information
Authors and Affiliations
Corresponding author
Editor information
Editors and Affiliations
Appendix A
Appendix A
This proof shows that the expected number of ‘expensive’ function evaluations of g (Eq. 2) is greater for LCB acquisition than for ZERO acquisition.
We first define Local Regret, which is the difference between a proposed function estimate and the actual function value. That is followed by Cumulative Regret, which a sum of local regrets. All other notation is defined in the main body of this paper. The following proof depends on a probability bound on the error estimate for g(xn) at an evaluation point xn due to Srinivas et al. [8]. With that bound, the general strategy is to calculate an upper bound for local regret, use that bound to determine the expected value of local regret, and then to calculate the expected number of ‘expensive’ function evaluations using the ZERO and LCB acquisition functions.
Definitions: Local regret \(r_{n}\) and Cumulative regret \(R_{N}\)
Definition: \(\beta_{n}\) Srinivas [8], Appendix A.1, Lemma 5.1, for constants Cn > 0 that satisfy \(\mathop \sum \nolimits_{n \ge 1} C_{n}^{ - 1} = 1\), and small \(\delta\):
Using the definition of local regret (Srinivas [8] Appendix A.1, Lemma 5.1), Equation (A4) provides upper and lower confidence bound for the GP “mu” term.
Then since \(\hat{x}\) is an optimal solution (so that \(g\left( {\hat{x}} \right) < g\left( {x_{n} } \right)\)), these upper and lower bounds apply respectively:
The general strategy in this proof is to estimate the maximum cumulative regret in each of these cases for LCB and ZERO acquisitions, and then calculate the expected difference between the two.
Proposition: ZERO acquisition converges faster than LCB acquisition in the case \(P\left( * \right) > 1 - \delta\) (Eq. A4).
Proof: First, from Eq. A4 with probability greater than \(1 - \delta\), the Local Regret calculation proceeds as in Eq. A6. The first line uses Eq. A4 with \(x = \hat{x}\), the second line uses Eq. A5 and the third line uses Eq. A4 with \(x = x_{n}\). The fourth line uses an upper bound: \(\beta_{n} = \mathop {\max }\limits_{n} \left( {\beta_{n} } \right)\).
Now consider the Local Regret in the cases of LCB and ZERO acquisition. ZERO acquisition is always non-negative but LCB acquisition can be negative. So we partition the values of n into those which result in zero or positive LCB acquisition (set S) and those which result in negative LCB acquisition (the complement, set \(S^{\prime}\)). These sets are shown in Eq. A7.
For S, the evaluation points proposed are identical for the two acquisition functions, since they both correspond to the same minimum. Therefore, using superscripts to denote the regrets for the two acquisition functions, the following equality applies.
For \(S^{\prime}\), ZERO acquisition returns a proposal that corresponds to a zero of the acquisition function, whereas the equivalent for LCB acquisition is negative, and we introduce a term \(\phi_{n}\) to account for the difference from zero (Eq. A9).
This leads to the following expressions for the two regrets, using Eq. A6.
Equation (A11) shows the partitioning the Cumulative Regret between sets S and \(S^{\prime}\).
Then, Eq. A12 show the maximum Cumulative Regret for ZERO and LCB acquisitions.
Equations A12 then imply that the inequality in Eq. A13.
Equation (A13) is a strong indication that ZERO acquisition leads to faster convergence than LCB acquisition, since it applies with high probability \(1 - \delta\). This completes the proof.
Rights and permissions
Copyright information
© 2021 The Author(s), under exclusive license to Springer Nature Switzerland AG
About this paper
Cite this paper
Mitic, P. (2021). Bayesian Optimization for Reverse Stress Testing. In: Vasant, P., Zelinka, I., Weber, GW. (eds) Intelligent Computing and Optimization. ICO 2020. Advances in Intelligent Systems and Computing, vol 1324. Springer, Cham. https://doi.org/10.1007/978-3-030-68154-8_17
Download citation
DOI: https://doi.org/10.1007/978-3-030-68154-8_17
Published:
Publisher Name: Springer, Cham
Print ISBN: 978-3-030-68153-1
Online ISBN: 978-3-030-68154-8
eBook Packages: Intelligent Technologies and RoboticsIntelligent Technologies and Robotics (R0)