PROJECT TITLE:

Efficient Motif Discovery for Large-scale Time Series in Healthcare - 2015

ABSTRACT:

Analyzing time series information will reveal the temporal behavior of the underlying mechanism manufacturing the information. Time series motifs, which are similar subsequences or frequently occurring patterns, have significant meanings for researchers especially in medical domain. With the quick growth of time series knowledge, ancient strategies for motif discovery are inefficient and not applicable to giant-scale knowledge. This work proposes an efficient Motif Discovery method for Large-scale time series (MDLats). By computing normal motifs, MDLats eliminates a majority of redundant computation within the related arts and reuses existing data to the maximum. All the motif varieties and subsequences are generated for subsequent analysis and classification. Our system is implemented on a Hadoop platform and deployed during a hospital for clinical electrocardiography classification. The experiments on real-world healthcare data show that MDLats outperform the state-of-the-art strategies even in large time series.