行政公告
  • 行政公告
4 月 28 日 (週五) 4 月份 WETA 研討會訊息
活動起日:2017-04-18 
發佈日期:2017-04-18 
瀏覽數:256  2017-04-18 更新

各位學界的朋友,大家好:

 

國立臺灣大學計量理論與應用研究中心 (CRETA)、國立臺灣大學財務金融學系及臺灣經濟計量學會 (TES) 將共同舉辦 4 月份WETA @TES4 月份 WETA 研討會相關資訊如下。

 

2017 4 月份 WETA 研討會】

日期:2017 4 28 (週五) 下午2:00~5:00

地點:國立臺灣大學管理學院二號館三樓 304 教室

講者:顏佐榕教授(中央研究院統計科學研究所)

演講主題:

(1) Estimating Links of a Network from Time to Event Data

(2) Solving Fused Group Lasso Problems via Block Splitting Algorithms

 

講題摘要:

(1)

In this paper we develop a statistical method for identifying links of a network from time to event data. This method models the hazard function of a node conditional on event time of other nodes, parameterizing the conditional hazard function with the links of the network. It then estimates the hazard function by maximizing a pseudo partial likelihood function with parameters subject to a user-specified penalty function and additional constraints. To make such estimation robust, it adopts a pre-specified risk control on the number of false discovered links by using the Stability Selection method. Simulation study shows that under this hybrid procedure, the number of false discovered links is tightly controlled while the true links are well recovered. We apply our method to estimate a political cohesion network that drives donation behavior of 146 firms from the data collected during the 2008 Taiwanese legislative election. The results show that firms affiliated with elite organizations or firms of monopoly are more likely to diffuse donation behavior. In contrast, firms belonging to technology industry are more likely to act independently on donation.

 

Keywords: Hazard network models; Right-censored data; Partial likelihood function; Stability Selection; Political cohesion networks.

 

(2)

In this paper we propose a distributed optimization-based method for solving the fused group lasso problem, in which the penalty function is a sum of Euclidean distances between pairs of parameter vectors. As a result of that, the penalty function is not separable in terms of these parameter vectors. To make the penalty function separable, one common way is to introduce a set of auxiliary variables that represent the differences between pairs of parameter vectors. This representation can be seen as a linear operator on the joint vector of the parameter vectors, and the resulting augmented Lagrangian will have a coupling quadratic term involving the linear representation. Even though the linear representation is separable in terms of the parameter vectors, the coupling quadratic term is not. To make the coupling quadratic term separable, we further introduce a set of equality constraints that connect each parameter vector to a group of paired auxiliary variables. With these newly introduced equality constraints, we are able to derive a modified augmented Lagrangian that is separable either in terms of the parameter vectors or in terms of the paired auxiliary variables. This separable property further facilitates us to solve the fused group lasso problem by developing an iterative algorithm with that most tasks can be carried out independently in parallel. We evaluate performance of the parallel algorithm by carrying out fused group lasso estimation for regression models using simulated data sets. Our results show that the parallel algorithm has a massive advantage over its non-parallel counterpart in terms of computational time and memory usage. In addition, with additional steps in each iteration, the parallel algorithm can obtain parameter values almost identical to those obtained by the non-parallel algorithm.

 

Keywords: Fused lasso; Group lasso; Scalability; Alternating direction method of multipliers; Block splitting algorithms.

 

講者介紹:

顏佐榕教授為倫敦帝國學院博士,目前任職於中央研究院統計科學研究所。研究專長為高維資料變數篩選、貝氏統計及社群網路資料分析,詳細資訊請見顏教授個人網站:http://www.stat.sinica.edu.tw/tjyen/index.html

 

會程安排:

下午 130 ~ 200報到

下午 200 ~ 320 First session

下午 320 ~ 340 Tea Break

下午 340 ~ 500 Second session

 

為方便場地安排及人數預估,欲參加WETA的朋友們,煩請事先報名。

當天講義將優先提供給報名者

報名網址:http://creta.org.tw/?news_2=201

報名費用:免費。

報名期限:4/27 () 13:30

歡迎各位踴躍參加!!

為方便臺灣經濟計量學會 (TES) 會員繳納 106 年度會費,本次活動開放現場繳納會費,亦歡迎大家介紹非會員朋友加入 TES。更多研討會資訊請見 TES 網站:http://www.tesociety.org.tw/main.php

 

此外,臺灣經濟計量學會新開設臉書粉絲專頁,歡迎各位舊雨新知按讚及追蹤我們各項活動的最新訊息!網址如下:

https://www.facebook.com/twtes/