主讲人：郭贵冰

郭贵冰，长聘教授，博导，辽宁省百千万人才，LibRec推荐系统开源库创始人，SCI二区期刊ECRA的编委。毕业于新加坡南洋理工大学，主要的研究兴趣包括推荐系统、自然语言处理、数据挖掘等。在相关研究领域已发表高水平学术论文90余篇，出版了1部学术专著《推荐系统进展：方法与技术》。根据Google学术统计，已发表论文共获得2900多次的学术引用。承担了国家自然科学基金面上项目、青年基金项目、教育部基本科研业务费优青培育项目、华为HIRP OPEN 2017基金项目、京东AI研究院“京东葡萄树--学者计划”等。

报告主题：Negative Sampling in Recommender Systems

日期：2022年1月6日星期四16:30-18:00

地点：L0316

Negative sampling has been widely adopted in recommender systems to provide a number of informative negative training examples, whereby better recommenders can be obtained. In this talk, I’ll first introduce the importance of negative sampling and briefly overview our previous early works on negative sampling. Then, I’ll talk about several recent works on negative sampling strategies designed by our research team. These works focus on how to effectively sample informative item pairs (rather than items) for collaborative filtering, user-item multi-hop path sampling for path-based recommenders, and item set sampling for GAN-based recommenders.