报告人：饶楠（Claremont Graduate university）

报告题目：Clustering Analysis on Stochastic Processes

报告摘要：Clustering involves partitioning data into multiple clusters, where any two objects in a cluster are close enough under some “distance”. Classical clustering techniques are designed for data points described by a finite number of features. However, there is growing interest in discovering patterns in data sampled from time series, which have an infinite number of features. Such time series are found in diverse areas ranging from biological and medical research to finance. Existing clustering methods fail on such problems.In this talk, I will introduce a new unsupervised learning problem: clustering wide-sense stationary ergodic stochastic processes. Covariance-based dissimilarity measures and consistent algorithms are designed for clustering offline and online data settings, respectively.

报告人简介：饶楠博士，本科毕业于Arizona大学，之后在Claremont Graduate University 获得金融工程硕士学位。她在博士阶段研究的课题是时间序列的无监督学习。最近的工作是关于随机过程的聚类分析，以及多重分数维布朗运动的一些问题。

报告时间：2018年6月4日（星期一）下午15:00-16:00

报告地点：科技楼南702