Timing the Decision Support for Real-World Many-Objective Optimization Problems

Abstract

Lately, there is growing emphasis on improving the scalability of multi-objective evolutionary algorithms (MOEAs) so that manyobjective problems (characterized by more than three objectives) can be effectively dealt with. Alternatively, the utility of integrating decision maker's (DM's) preferences into the optimization process so as to target some most preferred solutions by the DM (instead of the whole Paretooptimal front), is also being increasingly recognized. The authors here, have earlier argued that despite the promises in the latter approach, its practical utility may be impaired by the lack of-objectivity, repeatability, consistency, and coherence in the DM's preferences. To counter this, the authors have also earlier proposed a machine learning based decision support framework to reveal the preference-structure of objectives. Notably, the revealed preference-structure may be sensitive to the timing of application of this framework along an MOEA run. In this paper the authors counter this limitation, by integrating a termination criterion with an MOEA run, towards determining the appropriate timing for application of the machine learning based framework. Results based on three real-world many-objective problems considered in this paper, highlight the utility of the proposed integration towards an objective, repeatable, consistent, and coherent decision support for many-objective problems.

Benchmarking Real-World Many-Objective Problems: A Problem Suite With Baseline Results

“…However, the performance evaluation of most MOEAs is conducted using artificial test problems. Although these artificial test problems successfully encompass numerous challenging problem characteristics, it is still uncertain whether they can effectively replicate the complexity encountered in real-world applications [32]. The significant concern with numerically defined test problems is that they do not capture the complicated details of complex real-world problems and often require implementing nonlinear computer programs or black-box simulations.…”

“…1) First, due to agreement policies, the companies don't allow to reveal the mathematical formulations and details of the problems to safeguard their trade secrets [32]. 2) Next, implementing the problem requires some proprietary software for conducting simulations [35].…”

“…2) Next, implementing the problem requires some proprietary software for conducting simulations [35]. 3) Finally, individuals may refrain from disclosing their real-world problems publicly due to a perceived lack of advantages [32].…”

Many-Objective Real-World Engineering Problems: A Comparative Study of State-of-the-Art Algorithms

“…Most optimization algorithms developed for MaOPs have been tested on well-defined test problems to assess their efficiency and effectiveness. These test problems comprise a benchmark suite designed to reflect challenging problem characteristics, including but not limited to multimodality, discreteness, nonconvexity, deception, nonuniformity, isolated optima, non-separability, and scalability of the number of objectives as well as decision variables [34]. Although these benchmarks capture various challenging characteristics, whether they can capture the complexity observed in real-world applications remains to be seen [35].…”

“…Many comparative studies in the past analyzed the performance of optimization algorithms [37]. However, these were often restricted to artificial test problems, and only a few studies focused on using MOEAs for real-world applications [34]- [36]. Unfortunately, the existing studies on real-world applications are confined to just a few approaches to solving two or three real-world problems.…”