國立政治大學心理學系109-2學術研討會
講者：李玉麟 副教授（Weill Cornell Medical College, Department of Psychiatry）
講題：Bayesian Nonparametrics in Cognitive Psychology
時間：2021年05月17日（星期一）10：10-12：00
地點：國立政治大學研究大樓270105室
※本次演講為線上遠距形式，請於演講日使用google meet，輸入代碼為fyz-nsht-rdc
演講摘要：
2011年，MIT教授Josh Tenenbaum在Science發表了一篇影響深遠的論文，其中提到許多長久以來在認知科學上難以解決的問題都可以用Bayesian nonparametric (BNP) 的統計方法來找到合理的答案。但是這論文裡面有許多抽象又難懂的概念，讀了以後可能會更困惑。本次演講目的就是協助大眾解讀這些抽象的概念，像是"Chinese Restaurant Process"(CRP)。希望能解釋CRP，而且解釋CRP在認知心理學領域的功用，讓CRP對大眾而言不再莫測高深，進而再讀Tenenbaum的論文會有新的瞭解，借此能有不一樣的新想法和新發現。
講者簡介：
EDUCATIONAL BACKGROUND
|
Degrees |
Institution Name |
|
Bachelor of Science, Psychology |
National ChengChi University |
|
Master of Philosophy, Cognitive Psychology |
Columbia University |
|
Doctor of Philosophy, Cognitive Psychology |
Columbia University |
PROFESSIONAL POSITIONS AND EMPLOYMENT
|
Title |
Institution Name |
|
Research Assistant Professor |
University of Pennsylvania Medical School |
|
Associate Member |
Memorial Sloan-Kettering Cancer Center |
|
Associate Professor |
Psychology Research Department of Psychiatry, Weill Cornell Medical College |
|
Associate Attending Behavioral Scientist |
Memorial Hospital for Cancer and Allied Diseases, Department of Psychiatry & Behavioral Sciences, Behavioral Sciences Service; Department of Epidemiology & Biostatistics, Biostatistics Service |
Current Research Interests
Dr. Li is a behavioral statistician with a joint appointment in the Department of Psychiatry & Behavioral Sciences and the Department of Epidemiology and Biostatistics. His expertise is in Bayesian Item Response Theory and Bayesian Multilevel Modeling. His current research focuses on Bayesian Nonparametric methods in psychology as well as Patient-Reported Outcomes data analysis using Machine Learning analytics, including Dirichlet Process Mixture Modeling on psychological symptom clusters for patients with advanced/terminal cancer near the end of life and Latent Dirichlet Allocation to score patients’ quality of life. Clinicians are no longer restricted by conventional QOL surveys which rely on fixed (and often lengthy) surveys that routinely neglect each patient’s idiosyncratic goals and priorities. Patients’ goals can now be incorporated to enhance conventional data from fixed-length QOL measures. Dr. Li also works on making these revolutionary ideas accessible to behavioral scientists in his tutorials on Bayesian psychometric methods.
