A Survey of Unsupervised Pattern Discovery in Time Series

PhD Qualifying Examination


Title: "A Survey of Unsupervised Pattern Discovery in Time Series"

by

Mr. Fengchao PENG


Abstract:

Time series clustering is a hot topic in data mining. It is widely used in 
nance, bioin- formatics, sensor network and many other areas. In this survey, 
we provide an overview of recent progress in time series clustering, focusing 
on some bottleneck problems such as distance computation cost, interpretability 
and online discovery. Researchers pro- pose many novel methods to solve these 
problems. Moreover, some researchers also propose new denitions on time series 
pattern discovery, among which motif discovery and shapelet discovery are two 
prevalent topics. Motif discovery focuses on nding frequent or nearest 
patterns. Shapelet discovery is rather novel and attracting more and more 
attention. Methods on these two problems share many ideas with time series 
clustering, but these two problems have more detailed emphasis on the 
properties of results. Since they are closely related to clustering, we will 
also discuss them in this survey.


Date:			Friday, 8 April 2016

Time:                  	3:30pm - 5:30pm

Venue:                  Room 2612A
                         Lifts 31/32

Committee Members:	Prof. Lionel Ni (Supervisor)
 			Dr. Qiong Luo (Supervisor)
 			Dr. Lei Chen (Chairperson)
 			Dr. Ke Yi


**** ALL are Welcome ****