您当前的位置:首页 >科学研究 > 学术交流

学术报告六十八:Exploration versus exploitation in statistical learning

来源:数学与统计学院     作者:     时间:2018/7/10 11:08:13  0次

数学与统计院学术报告[2018] 068

(高水平大学建设系列报告187)

报告题目:  Exploration versus exploitation in statistical learning

报告人:王熙逵 教授(加拿大 曼尼托巴大学

报告时间:71110:30-11:30

报告地点:科技楼501               

报告内容:

The bandit process is a statistical model of sequential learning and decision making. Sequential observations are made from several statistics populations, some or all of which have unknown statistical distributions. Each observation has two consequences. On one hand, it provides information about the distribution of the selected population. This is exploration and benefits later decisions because it helps reduce uncertainty when the population distribution is unknown. On the other hand, each observation defines a reward or cost. This is exploitation. In reality, there is a tradeoff between exploration and exploitation in virtually every statistical learning problem. We discuss the models, methods and applications of bandit processes. 

报告人简历:

 

  王熙逵教授自 1988年一共发表学术论文 40余篇,几乎全部发表在国际 SCI学术刊物上,包括统计学的前沿学术杂志,如 Biometrical Journal, Annals of the Institute of Statistical Mathematics, Journal of Statistical Planning and Inference, North American Actuarial Journal, 等等。至今为止,在国际学术会议上已经做过 30余次特邀和大会主题报告。 英国皇家统计协会会士(Fellow of the Royal Statistical Society),国际数理统计协会(Institute of Mathematical Statistics) 和泛华统计协会(International Chinese Statistical Association) 终身会员, 加拿大统计协会(Statistical Society of Canada) 会员和职业统计师(Professional Statistician)

数学与统计学院

2018710

分享到: 0

数学与统计学院研究生工作奖励措施

2016年深圳大学数学与统计学院暨暑期夏令营招生指南