Find robust solutions over time by two-layer multi-objective optimization method

A Simple Yet Effective Approach to Robust Optimization Over Time

“…Benchmark 1 with λ = 1 is taken from [8], Benchmark 1 with λ = 0 from [11] and Benchmark 2 from [2]. Note that [11] compared himself with the results from [8], [9], [10] and showed that their results are superior. For Benchmark 2 we considered only the random movement which in [2] was denoted as T P 13 .…”

“…In this section, we describe the moving peak benchmark commonly used in the ROOT literature. It is based on [15] and appeared in many papers [2], [7], [8], [9], [10], [11], [12], [13]. However, to the best of our knowledge, no complete and proper description was given in any of these papers.…”

Robust Optimization Over Time by Learning Problem Space Characteristics

“…Also, two different benchmark problems were proposed. In [10], a new two-layer multi-objective method was proposed to find robust solutions that can maximize both survival time and average fitness. In [11], another multiobjective method was proposed to minimize switching cost and maximize survival time.…”

“…However, removing the approximator and predictor from algorithms that work based on future fitness values of solutions clearly is a substantial simplification and the performance on a real-world problem where future fitness values are not available may be very different. Overall, for solving real-world problems, almost all the current ROOT methods [6], [8]- [10] and [11] need to use approximation and prediction methods based on time series [12].…”

“…In this paper, and similar to ROOT papers in [6], [8], [10], [11], [34], we use the standard baseline function of MPB as follows:…”