光华讲坛——社会名流与企业家论坛第6588期
主题:Introduction to Multi-armed Bandits 多臂老虎机导论(系列讲座)
主讲人:伊利诺伊大学芝加哥分校 周文心副教授
主持人:西南财经大学统计学院 常晋源教授
时 间:6月26日 14:00-17:00 6月27日 14:00-17:00 7月2日 14:00-17:00 7月3日 14:00-17:00
举办地点:西南财经大学光华校区光华裙楼2303教室
主办单位:数据科学与商业智能联合实验室 统计学院 科研处
主讲人简介:
Wen-Xin Zhou is an Associate Professor in the Department of Information and Decision Sciences at the College of Business Administration, University of Illinois at Chicago. From 2017 to 2023, he was a faculty member in the Department of Mathematics at the University of California, San Diego. His research interests include high-dimensional statistics, robust learning for heavy-tailed data, nonparametric statistics, neural networks and deep learning, quantile regression methods and beyond.
周文心,伊利诺伊大学芝加哥分校工商管理学院信息与决策科学系副教授。2017 年至 2023 年,他在加州大学圣地亚哥分校数学系任教。他的研究兴趣包括高维统计、重尾数据的稳健学习、非参数统计、神经网络和深度学习、量化回归方法等。
内容简介:
Multi-armed bandits are a simple but powerful framework for algorithms that make decisions over time under uncertainty. An enormous body of work has accumulated over the years, covered in several books and surveys. In these talks, we provide an introductory treatment of the subject. Each chapter tackles a particular line of work, offering a self-contained, teachable technical introduction and a brief review of further developments.
To begin with, we discuss a spectrum of decision-making problems, online learning, and prediction. Then, we describe some of the fundamental algorithms for multi-armed bandits. The second chapter concerns contextual bandits, focusing on stochastic linear bandits. In chapter three, we consider the stochastic sparse linear bandit problem, where only a sparse subset of context features affects the expected reward function. We will review some representative and very recent works in this direction and discuss some open questions.
多臂老虎机是一个简单但强大的算法框架,用于在不确定性下随时间做出决策。多年来,关于这一主题的大量研究成果积累了下来,并已在多本书籍和综述中有所涵盖。在本系列讲座中,我们将对该主题进行介绍。每一章都会讨论一个特定的研究方向,提供一个自成体系的技术介绍,并简要回顾相关的最新进展。
首先,将讨论一系列决策问题、在线学习和预测。然后,将讲解一些多臂老虎机的基本算法。第二章将关注上下文老虎机,特别是随机线性老虎机。第三章,将讲解随机稀疏线性老虎机问题,其中只有稀疏的上下文特征子集会影响期望奖励函数。最后,将回顾该方向上一些具有代表性和最新的研究工作,并讨论一些未解决的问题。