主講人：陳瑞彬 教授 (高雄大學統計學研究所)

主題：Surrogate Assistant Approaches for Searching Effective Points in Computer Experiments

Abstract：
In computer experiments, true response functions are usually assumed to be unknown, and the cost for obtaining a function response is very expensive. Thus instead of identifying true response surfaces directly, we would like to approximate response surfaces by inexpensive surrogates which are built only by using few explored experimental points. Surrogate assistant methods work successfully in fitting response surfaces and processing optimization. In this talk, we would introduce some useful surrogate construction methods, and we would then propose several surrogate assistant approaches for searching effective points and finding local optimal points.

時間：2008 年 12 月 15 日13:15~14:10

地點：新生大樓202 室

茶會：14:20於舊數館201室