Welcome Tp CRETA! Contact Us at : +886.2.3366.1072

Welcome To CRETA

國立臺灣大學計量理論與應用研究中心 - CRETA

學術活動

CRETA Seminar

2024 年 4 月 26 日 CRETA Seminar - Multiple-Instance Learning with Heterogeneous Instances

訊息標題:

2024 年 4 月 26 日 CRETA Seminar - Multiple-Instance Learning with Heterogeneous Instances

簡介摘要:

國立臺灣大學計量理論與應用研究中心 (CRETA) 及臺灣經濟計量學會 (TES) 將於 2024 年 4 月 26 日舉辦 CRETA Seminar。相關資訊如下:

 

 4  26  CRETA Seminar

日期:2024 年 4 月 26 日 (週五) 下午 2:00~3:30

地點:國立臺灣大學管理學院二號館 103 教室

講者:

楊鈞澔教授 (國立臺灣大學統計與數據科學研究所)

演講主題:Multiple-Instance Learning with Heterogeneous Instances

講題摘要:

Multiple-instance learning (MIL) is a weakly supervised learning problem, where a single class label is assigned to a bag of instances, and the instance labels are not directly observed. The standard assumption for the relationship between bags and instances is that a bag has a positive label if at least one instance is positive. This assumption allows us to infer the instance-level information through the bag-level observations. MIL has received a lot of attention in the past decade due to its wide applications in biology, chemistry, computer vision, etc. In this talk, I will briefly introduce some of my research works in MIL including (i) the multiple-instance logistic regression (MILR) model, and (ii) the heterogeneous-instance logistic regression (HILoR) model, which is a modification of MILR to accommodate the heterogeneity in predictors among different instances. 

As an application of the proposed HILoR model, I will use the multiple-criterion diagnoses for mild cognitive impairment (MCI) and Alzheimer’s disease (AD) as an example. MCI is a prodromal stage of AD that causes a significant burden in caregiving and medical costs. Clinically, the diagnosis of MCI is determined by the impairment statuses of five cognitive domains. If one of these cognitive domains is impaired, the patient is diagnosed with MCI, and if two out of the five domains are impaired, the patient is diagnosed with AD. The proposed model is validated in terms of its estimation accuracy, latent status prediction, and robustness via extensive simulation studies and the National Alzheimer’s Coordinating Center-Uniform Data Set.

 

備註事項:

欲參加CRETA Seminar的朋友們,煩請事先報名  

報名期限:2024/4/25 (四)  中午 12:00

報名網址:https://docs.google.com/forms/d/1TefqTA6V-433xliFiVcSeD_OtI8Zov4kWnxM0REmqbU/edit

歡迎踴躍報名參加!