Exclusive Content & Downloads from ASQ

Comment- Properties and Practicalities of the Expected Quantile Improvement

Summary: [This abstract is based on the authors' abstract.] The expected quantile improvement (EQI) of Picheny et al. (Victor Picheny, David Ginsbourger, Yann Richet, and Gregory Caplin, Quantile-Based Optimization of Noisy Computer Experiments with Tunable Precision, Technometrics, 55-1, pages 2-13) is an extension of the expected improvement (EI) sequential sampling criterion for selecting experiments that can be performed at varying levels of accuracy (yielding varying degrees of noise). It is an elegant formulation that reduces to the original EI in the absence of noise. This comment first examines the properties of the EQI, following on from the discussion in Picheny et al. The effect of each control parameter, which essentially determines to what extent the EQI departs from the EI, is briefly investigated using a one-dimensional test function. Following this, the EQI is considered in the context of solving practical engineering problems; the article looks at the assumptions made about the convergence properties of simulators, which are key to the success of the method, and at implementation issues, including the efficient application of computational resources.

Anyone with a subscription, including Site and Enterprise members, can access this article.

Other Ways to Access content:

Join ASQ

Join ASQ as a Full member. Enjoy all the ASQ member benefits including access to many online articles.

  • Topics: Statistics
  • Keywords: Stochastic models, Kriging, Noise, Precision, Accuracy and precision, Design of experiments (DOE), Statistical experimental design (SED), Convergence
  • Author: Forrester, Alexander I. J.
  • Journal: Technometrics