Topic:

A Joint Interactive Modeling Approach to Fashion Outfit Recommender Systems

Time&Date: 

10:30 am - 12:00 pm, July 9, 2024 (Tuesday)

Venue

Room 619, Teaching A Building

Zoom Link: 
https:

https://cuhk-edu-cn.zoom.us/j/3985407949?pwd=QnZJMHU3SDUwaFdtWTF6N3RWcGlMdz09

Meeting ID: 398 540 7949

Passcode: 779898

Speaker:

Dr. Xuan Bi (University of Minnesota)

With the advancement of machine learning and artificial intelligence technologies, recommender systems have been widely adopted across various fields. One notable application is in the fashion domain, where stylistic compatibility is often required. Typically, this necessitates data from fashion images or text descriptions, which can limit the ability to personalize recommendations. In this work, we propose a joint modeling framework that can learn stylistic compatibility without relying on image or text data. Our method preserves the ability to provide personalized recommendations while being able to recommend both individual articles and outfits. We evaluate the proposed method through multiple simulation studies, an offline fashion data analysis, and an online experiment on Prolific. All results consistently demonstrate the effectiveness of our method.

Biography:

Dr. Xuan Bi is an Assistant Professor of Information and Decision Sciences at the Carlson School of Management at the University of Minnesota. His research revolves around personalization and trustworthy machine learning, with a focus on recommender systems and data privacy. His works have been published in leading academic journals, including Management Science, Information Systems Research, Journal of the American Statistical Association, Annals of Statistics, Journal of Machine Learning Research, INFORMS Journal on Computing, and Journal of Econometrics, and he serves as Associate Editor for the Journal of the American Statistical Association.