Curse and Blessing of Uncertainty in Evolutionary Algorithm Using Approximation

Abstract

Evolutionary frameworks that employ approximation models or surrogates for solving optimization problems with computationally expensive fitness functions may be referred as Surrogate-Assisted Evolutionary Algorithms (SAEA). In this paper, we present a study on the effects of uncertainty in the surrogate on SAEA. In particular, we focus on both the 'curse of uncertainty' and 'blessing of uncertainty' on evolutionary search, a notion borrowed from 'curse and blessing of dimensionality' in [1]. Here, the 'curse of uncertainty' refers to impairments due to the errors in the approximation. The 'blessing of uncertainty' is less explicitly discussed in the literature, but refers to the benefits of approximation errors on evolutionary search. Empirical studies suggest that approximation errors lead to convergence at false global optima, but prove to be beneficial in some cases.

Meta-Heuristics for Engineering Optimisation - Applications to Metal Forming Processes

“…However, the true best result after performing optimisation is obtained from the use of hybrid Kriging-support vector regression (KG-SVR). This situation is called the blessing and curse of uncertainties [14]. Figure 5 shows comparison between the design solutions from the conventional circular wire drawing, NCD, optimum NCD using KG-SVR, and hybrid Polynomial response surface (PRS), Radial basis function (RBF), KG and SVR.…”

Transfer optimization in complex engineering design

“…In the first step of this analysis, the Pearson correlation coefficient is calcu- The results confirms the expectations that as the source-target similarities increase, there is a corresponding monotonic improvement in the overall optimization performance. To further verify the assertion that this improvement is a consequence of improved modelling accuracy (due to knowledge transfer), rather than, for example, the blessings of uncertainty [151], the generalization error (RMSE) of the TS-GP model is compared with GP. Specifically, as shown in Note that GP subscript in the "Generalization Performance Gain" function refers to the no-transfer case.…”