Offline Data-Driven Evolutionary Optimization Using Selective Surrogate Ensembles

Abstract

In solving many real-world optimization problems, neither mathematical functions nor numerical simulations are available for evaluating the quality of candidate solutions. Instead, surrogate models must be built based on historical data to approximate the objective functions and no new data will be available during the optimization process. Such problems are known as offline data-driven optimization problems. Since the surrogate models solely depend on the given historical data, the optimization algorithm is able to search only in a very limited decision space during offline data-driven optimization. This paper proposes a new offline data-driven evolutionary algorithm to make the full use of the offline data to guide the search. To this end, a surrogate management strategy based on ensemble learning techniques developed in machine learning is adopted, which builds a large number of surrogate models before optimization and adaptively selects a small yet diverse subset of them during the optimization to achieve the best local approximation accuracy and reduce the computational complexity. Our experimental results on the benchmark problems and a transonic airfoil design example show that the proposed algorithm is able to handle offline data-driven optimization problems with up to 100 decision variables.

Off-line Data-driven Multi-objective Optimization: Knowledge Transfer between Surrogates and Generation of Final Solutions

“…Off-line data-driven optimization problems widely exist in the real world [11], [36], such as trauma systems design [6], performance optimization of fused magnesium furnaces [7], and operational indices optimization of beneficiation processes [8], in which the objective functions cannot be directly calculated using mathematical equations and only data are available for fitness evaluations [11], [36]. Off-line data-driven optimization starts with a certain amount of collected data, which are used to construct surrogates for searching optimal solutions [9]. During the optimization, surrogate models can be updated to improve the search efficiency either by using generated synthetic data from other surrogates, re-using knowledge collected from the optimization, or newly collected data that are not under the control of the optimizer.…”

“…During the optimization, surrogate models can be updated to improve the search efficiency either by using generated synthetic data from other surrogates, re-using knowledge collected from the optimization, or newly collected data that are not under the control of the optimizer. An example of model management techniques using information during the optimization can be found in [9], which adaptively selects a subset of the base learners of an ensemble at each iteration according to the location of the best individual. Once the surrogate-assisted optimization is completed, the best solution (set) will be implemented to solve the real-world problem.…”

“…In the on-line data-driven optimization, a small number of candidates can be validated during the optimization with the real objective functions (obtained by either performing numerical simulations or experiments), and the newly generated data can also be used to update the surrogates. By contrast, in the off-line datadriven optimization, no new data can be generated and only the historical data can be exploited to manage the surrogates [7], [9]. Most data-driven optimization relies on surrogate models to assist the optimizer to guide the search [11]- [13].…”

“…In [7], a low-order PR model and a GP model are constructed, where the low-order PR is used as the real fitness function to produce new data during search process for model management, while the GP model is employed as the surrogate to assist the evolutionary search. In [9], a large number of base learners are built off-line and then a subset of these base learners are adaptively selected to make sure that the surrogate is able to best approximate the local fitness landscape. Note that the algorithm reported in [6] is developed for optimization driven by a large amount of data, which is meant for reducing the computation time by using less data by means of adaptive clustering.…”

Symbiotic Organism Search Algorithm with Multi-Group Quantum-Behavior Communication Scheme Applied in Wireless Sensor Networks

“…Although the objective function has multiple constraints in the objective world, the performance of the objective function of the algorithm under unconstrained conditions plays a fundamental role in various optimization applications, which will then affect the performance and direction of the algorithm in a constrained objective function. Computational intelligence (CI) [6][7][8][9][10] not only deals with complex problems in real life (most notably the objective functions of uncertain or noisy problems [11]), it can also give calculation methods and solutions to solve these optimization problems. Evolutionary computation (EC) [12][13][14][15] is a branch of CI that provides an optimization method with evolutionary ideas.…”