Computational Technology Resources - CCP

Keywords: non-parametric stochastic subset optimization, optimization under uncertainty, kernel density estimation, adaptive kernel sampling density, stochastic sampling.

Summary

Non-parametric stochastic subset optimization (NP-SSO) was recently proposed for design problems utilizing the system reliability as the objective function. In this paper it is extended to general optimization under uncertainty problems (i.e. adopting a general probabilistic measure as the objective function) and it is also coupled with various computational advances to improve its non-parametric characteristics and its numerical efficiency. The algorithm relies on simulation of samples of the design variables from an auxiliary probability density function, and uses the information in these samples to provide an efficient approximation of the objective function through kernel density estimation (KDE). Candidate points for the global minimum are then identified based on this approximation whereas an iterative approach is established for improved computational efficiency. To deal with the boundary correction for the KDE in complex domains (defining the search space at each iteration), a multivariate boundary KDE based on local linear estimation is proposed here. Also, a non-parametric characterization of the optimal subset using a framework based on support vector machine is proposed to facilitate efficient simulation of samples and estimation of parameters required for the boundary correction within the proposed iterative scheme. To further improve the efficiency of the stochastic sampling stage, required within NP-SSO to generate the required samples from the design variables, an adaptive kernel sampling density is proposed; samples from the previous iteration are used to seamlessly formulate and optimize efficient proposal densities for this purpose. The advances proposed here ultimately contribute greatly to the efficiency of NP-SSO, facilitating a more versatile characterization of the search space at each iteration as well as improving the sampling efficiency.

purchase the full-text of this paper (price £20)

go to the previous paper
go to the next paper
return to the table of contents
return to the book description
purchase this book (price £42 +P&P)

	Computational & Technology Resources an online resource for computational, engineering & technology publications
	not logged in - login
Front Page Browse CCP CSETS CTR IJRT Other Authors Search Purchase Guide FAQ Contact us	Civil-Comp Proceedings ISSN 1759-3433 CCP: 103 PROCEEDINGS OF THE THIRD INTERNATIONAL CONFERENCE ON SOFT COMPUTING TECHNOLOGY IN CIVIL, STRUCTURAL AND ENVIRONMENTAL ENGINEERING Edited by: Y. Tsompanakis Paper 10 Non-Parametric Stochastic Subset Optimization for System Design Optimization under Uncertainty G. Jia and A.A. Taflanidis Department of Civil & Environmental Engineering & Earth Sciences, University of Notre Dame, United States of America doi:10.4203/ccp.103.10 purchase the full-text of this paper Full Bibliographic Reference for this paper G. Jia, A.A. Taflanidis, "Non-Parametric Stochastic Subset Optimization for System Design Optimization under Uncertainty", in Y. Tsompanakis, (Editor), "Proceedings of the Third International Conference on Soft Computing Technology in Civil, Structural and Environmental Engineering", Civil-Comp Press, Stirlingshire, UK, Paper 10, 2013. doi:10.4203/ccp.103.10 Keywords: non-parametric stochastic subset optimization, optimization under uncertainty, kernel density estimation, adaptive kernel sampling density, stochastic sampling. Summary Non-parametric stochastic subset optimization (NP-SSO) was recently proposed for design problems utilizing the system reliability as the objective function. In this paper it is extended to general optimization under uncertainty problems (i.e. adopting a general probabilistic measure as the objective function) and it is also coupled with various computational advances to improve its non-parametric characteristics and its numerical efficiency. The algorithm relies on simulation of samples of the design variables from an auxiliary probability density function, and uses the information in these samples to provide an efficient approximation of the objective function through kernel density estimation (KDE). Candidate points for the global minimum are then identified based on this approximation whereas an iterative approach is established for improved computational efficiency. To deal with the boundary correction for the KDE in complex domains (defining the search space at each iteration), a multivariate boundary KDE based on local linear estimation is proposed here. Also, a non-parametric characterization of the optimal subset using a framework based on support vector machine is proposed to facilitate efficient simulation of samples and estimation of parameters required for the boundary correction within the proposed iterative scheme. To further improve the efficiency of the stochastic sampling stage, required within NP-SSO to generate the required samples from the design variables, an adaptive kernel sampling density is proposed; samples from the previous iteration are used to seamlessly formulate and optimize efficient proposal densities for this purpose. The advances proposed here ultimately contribute greatly to the efficiency of NP-SSO, facilitating a more versatile characterization of the search space at each iteration as well as improving the sampling efficiency. purchase the full-text of this paper (price £20) go to the previous paper go to the next paper return to the table of contents return to the book description purchase this book (price £42 +P&P)
Back to top	©Civil-Comp Limited 2023 - terms & conditions