Fast Exact Computation of Expected HyperVolume Improvement

In multi-objective Bayesian optimization and surrogate-based evolutionary algorithms, Expected HyperVolume Improvement (EHVI) is widely used as the acquisition function to guide the search approaching the Pareto front. This paper focuses on the exact calculation of EHVI given a nondominated set, for which the existing exact algorithms are complex and can be inefficient for problems with more than three objectives. Integrating with different decomposition algorithms, we propose a new method for calculating the integral in each decomposed high-dimensional box in constant time. We develop three new exact EHVI calculation algorithms based on three region decomposition methods. The first grid-based algorithm has a complexity of $O(m\cdot n^m)$ with $n$ denoting the size of the nondominated set and $m$ the number of objectives. The Walking Fish Group (WFG)-based algorithm has a worst-case complexity of $O(m\cdot 2^n)$ but has a better average performance. These two can be applied for problems with any $m$. The third CLM-based algorithm is only for $m=3$ and asymptotically optimal with complexity $\Theta(n\log{n})$. Performance comparison results show that all our three algorithms are at least twice faster than the state-of-the-art algorithms with the same decomposition methods. When $m>3$, our WFG-based algorithm can be over $10^2$ faster than the corresponding existing algorithms. Our algorithm is demonstrated in an example involving efficient multi-objective material design with Bayesian optimization.

[1]  Peter Stone,et al.  Reinforcement learning , 2019, Scholarpedia.

[2]  Andrzej Jaszkiewicz,et al.  Improved quick hypervolume algorithm , 2016, Comput. Oper. Res..

[3]  Michael T. M. Emmerich,et al.  Single- and multiobjective evolutionary optimization assisted by Gaussian random field metamodels , 2006, IEEE Transactions on Evolutionary Computation.

[4]  K. Chaloner,et al.  Bayesian Experimental Design: A Review , 1995 .

[5]  Tom Fawcett,et al.  Data Science and its Relationship to Big Data and Data-Driven Decision Making , 2013, Big Data.

[6]  Tom Dhaene,et al.  Fast calculation of multiobjective probability of improvement and expected improvement criteria for Pareto optimization , 2014, J. Glob. Optim..

[7]  Erik Brynjolfsson,et al.  Big data: the management revolution. , 2012, Harvard business review.

[8]  Warren B. Powell,et al.  The Knowledge-Gradient Policy for Correlated Normal Beliefs , 2009, INFORMS J. Comput..

[9]  G. Kang,et al.  A micromechanical constitutive model for anisotropic cyclic deformation of super-elastic NiTi shape memory alloy single crystals , 2015 .

[10]  Hao Wang,et al.  A Multicriteria Generalization of Bayesian Global Optimization , 2016, Advances in Stochastic and Deterministic Global Optimization.

[11]  Shigeru Obayashi,et al.  A Study on Many-Objective Optimization Using the Kriging-Surrogate-Based Evolutionary Algorithm Maximizing Expected Hypervolume Improvement , 2015 .

[12]  Julien Bect,et al.  A Bayesian approach to constrained single- and multi-objective optimization , 2015, Journal of Global Optimization.

[13]  M. Emmerich,et al.  The computation of the expected improvement in dominated hypervolume of Pareto front approximations , 2008 .

[14]  Carlos M. Fonseca,et al.  Computing 3-D Expected Hypervolume Improvement and Related Integrals in Asymptotically Optimal Time , 2017, EMO.

[15]  Andrew G. Barto,et al.  Reinforcement learning , 1998 .

[16]  Dimitris C. Lagoudas,et al.  Predictive Modeling of the Constitutive Response of Precipitation Hardened Ni-Rich NiTi , 2017, Shape Memory and Superelasticity.

[17]  Victor Picheny,et al.  Multiobjective optimization using Gaussian process emulators via stepwise uncertainty reduction , 2013, Statistics and Computing.

[18]  Carlos M. Fonseca,et al.  Computing Hypervolume Contributions in Low Dimensions: Asymptotically Optimal Algorithm and Complexity Results , 2011, EMO.

[19]  Michael T. M. Emmerich,et al.  Faster Exact Algorithms for Computing Expected Hypervolume Improvement , 2015, EMO.

[20]  Shiwei Zhou,et al.  On design of multi-functional microstructural materials , 2012, Journal of Materials Science.

[21]  Nicola Beume,et al.  On the Complexity of Computing the Hypervolume Indicator , 2009, IEEE Transactions on Evolutionary Computation.

[22]  Lucas Bradstreet,et al.  A Fast Way of Calculating Exact Hypervolumes , 2012, IEEE Transactions on Evolutionary Computation.

[23]  Carlos M. Fonseca,et al.  A Box Decomposition Algorithm to Compute the Hypervolume Indicator , 2015, Comput. Oper. Res..

[24]  R. Lyndon While,et al.  A faster algorithm for calculating hypervolume , 2006, IEEE Transactions on Evolutionary Computation.