Multiresolution Motif Discovery in Time Series

Abstract

Time series motif discovery is an important problem with applications in a variety of areas that range from telecommunications to medicine. Several algorithms have been proposed to solve the problem. However, these algorithms heavily use expensive random disk accesses or assume the data can fit into main memory. They only consider motifs at a single resolution and are not suited to interactivity. In this work, we tackle the motif discovery problem as an approximate Top-K frequent subsequence discovery problem. We fully exploit state of the art iSAX representation multiresolution capability to obtain motifs at different resolutions. This property yields interactivity, allowing the user to navigate along the Top-K motifs structure. This permits a deeper understanding of the time series database. Further, we apply the Top-K space saving algorithm to our frequent subsequences approach. A scalable algorithm is obtained that is suitable for data stream like applications where small memory devices such as sensors are used. Our approach is scalable and disk-efficient since it only needs one single pass over the time series database. We provide empirical evidence of the validity of the algorithm in datasets from different areas that aim to represent practical applications.

Classifying Urban Sounds using Time Series Motifs

“…Castro el al. [6] have proposed a Multiresolution Motif Discovery in Time Series algorithm (MrMotif) based on the iSAX representation. This algorithm solves the motif discovery problem as an approximate Top-K frequent subsequence discovery problem.…”

A Haar Wavelet-based Multi-resolution Representation Method of Time Series Data

“…Using multi-resolution techniques to effectively visualize large time series is also applied in (Hao et al, 2007) where the proposed framework uses multiple resolution levels. In (Castro and Azevedo, 2010) the authors propose a method based on the multi-resolution property of iSAX (Shieh and Keogh, 2008), (Shieh and Keogh, 2009) to derive motifs at different resolutions. In (Lin et al, 2005) the authors propose a multi-resolution PAA (Keogh et al, 2000), (Yi and Faloutsos, 2000); a well-known time series dimensionality reduction technique, to achieve an algorithm for iterative clustering.…”

Visualizing Variable-Length Time Series Motifs

“…More specifically, a time series motif is a pattern that consists of two or more similar subsequences based on some distance threshold [18,26]. Since then, a great deal of work has been proposed for the discovery of time series motifs [3,4,5,18,20,21,22,23,25,26,29,30,31]. Figure 1 shows an example of a time series motif in an insect behavior dataset [23].…”

“…As we will demonstrate, not only can the inter-motif subsequences have variable lengths, the intra-motif subsequences also are not restricted to have identical length-a unique property that is desirable, but has not been seen in the literature. While the results produced by our algorithm are approximate solutions, it has been shown recently that in many applications, approximate solutions might be sufficient or even preferable due to efficiency [4].…”