ADAPTIVE AND ROBUST MULTI-TASK LEARNING

• 成果类型: Article
• 署名作者: Duan, Yaqi; Wang, Kaizheng
• 署名单位: New York University; Columbia University; Columbia University
• 刊物名称: ANNALS OF STATISTICS
• ISSN/ISSBN: 0090-5364
• DOI: 10.1214/23-AOS2319
• 发表日期: 2023
• 页码: 2015-2039
• 关键词: multiple tasks; algorithms; inference; BAYES; MODEL; RISK; PCA

摘要

We study the multitask learning problem that aims to simultaneously an-alyze multiple data sets collected from different sources and learn one model for each of them. We propose a family of adaptive methods that automatically utilize possible similarities among those tasks while carefully handling their differences. We derive sharp statistical guarantees for the methods and prove their robustness against outlier tasks. Numerical experiments on synthetic and real data sets demonstrate the efficacy of our new methods.