DMCA
Dynamic Mixture Models for Multiple Time Series
Cached
Download Links
| Citations: | 9 - 0 self |
BibTeX
@MISC{Wei_dynamicmixture,
author = {Xing Wei and Jimeng Sun and Xuerui Wang},
title = {Dynamic Mixture Models for Multiple Time Series},
year = {}
}
Share
 |
 |
 |
 |
||
OpenURL
Abstract
Traditional probabilistic mixture models such as Latent Dirichlet Allocation imply that data records (such as documents) are fully exchangeable. However, data are naturally collected along time, thus obey some order in time. In this paper, we present Dynamic Mixture Models (DMMs) for online pattern discovery in multiple time series. DMMs do not have the noticeable drawback of the SVD-based methods for data streams: negative values in hidden variables are often produced even with all nonnegative inputs. We apply DMM models to two real-world datasets, and achieve significantly better results with intuitive interpretation. 1
Keyphrases
dynamic mixture model multiple time series dmm model real-world datasets hidden variable data record latent dirichlet allocation negative value traditional probabilistic mixture model noticeable drawback data stream svd-based method nonnegative input intuitive interpretation online pattern discovery
