作者ant0314 (*螞蟻*)
看板NCTU-STAT98G
標題[公告] 統計所專題演講
時間Thu Sep 24 23:02:23 2009
國家理論科學研究中心學術演講
NCTS Industrial Statistics Research Group Seminar
Date: 11:00am~4:00pm, Sept. 25 (Fri.), 2009
(1) 11:00am-1:00pm, Group Meeting
----------------------------------------------------------------------------
(2) 1:00pm-2:30pm
Speaker: Prof. Shuen-Lin Jeng 鄭順林 (NCKU)
Topic: Reliability Models on Nanotechnology
Abstract: In wear, aging, and fatigue growth processes, continuous degradation models (CDMs) have been applied to assess device reliability for decades. On the other hand, in practical applications, discrete degradation processes (DDPs) have been also observed for years and are discussed again lately for leakage current of thin gate oxides in nano-technology. Here a DDP means that the occurrence of degradation event follows a certain discrete mechanism in a specified time interval and each discrete
degradation event leads to an increment of damage. Besides the two separate kinds of degradation types, mixed degradation processes (MDPs), which integrate both continuous and discrete degradation patterns, are observed recently in the measurements of electric properties of IC products. The first part of this talk will be the model building for MDPs. The second part will be a brief review of some reliability research on nanotechnology for open discussion.
----------------------------------------------------------------------------
(3) 2:30pm-4:00pm
Speaker: Prof. Jyh-Jen Horng 洪志真 (NCTU)
Topic: Profile Monitoring簡介
Abstract: 製程監控在品管上是很重要的一環。近年來統計品質管制 (SPC) 上的研究愈來愈多,範疇也愈來愈廣。由單變量的問題推廣至多變量的問題;由獨立性資料推廣至自我相關 (auto-correlated) 資料等。近年來有一個新的研究主題相當熱門: SPC for Profile Monitoring。一般而言,品質特性 (quality characteristic)
資料,無論是單變量或多變量,都是一個變數。然而對某些製程而言,一個產品之品質特性是由反應變數和一或多個解釋變數間之關係(relationship)來界定,而不再是一般之變數。因此一個產品之品質特性乃以一條曲線或是一個曲面或其離散化(discretized) 後之資料型式來呈現,稱之為profile (縱斷面或剖面)。例如Kang and Albin (2000)文中有一個例子為苯丙氨酸甲脂 (aspartame, 一種低熱量人工甜味劑)在每公升的水中所溶解的量與水溫度之關係,每個樣本即為一條曲線。此種形式之資料又稱為曲線資料 (curve
data),處理這種資料的統計模型很多,可分為參數迴歸和半參數/無參數迴歸,其中函數資料分析 (functional data analysis),在半參數/無參數迴歸的研究領域內是目前相當受到重視且還有許多研究空間的研究課題。本人不揣淺陋介紹大家一些我所知道的Profile Monitoring上的研究,希望藉此拋磚引玉,引起大家共同合作研究的興趣。
Place:
Lecture Room A of National Center for Theoretical Sciences, 4th Floor, The 3rd General Building, National Tsing Hua University
--
※ 發信站: 批踢踢實業坊(ptt.cc)
◆ From: 140.113.252.20
※ 編輯: ant0314 來自: 140.113.252.20 (09/24 23:08)