Abstract | ||
---|---|---|
The technique of searching for similar patterns among time series data is very important in a wide range of applications. Among them, the time scaling searching is a hard problem that only a few works have tackled. By combining the advantages of a natural time scaling transformation and the dynamic time warping method, we propose a similarity measure that is more suitable for time scaling searching than any existing one. We then explain how to calculate the proposed segment-wise time warping (STW) distance using dynamic programming. In addition, we discuss the lower bound technique of STW distance and the corresponding index method. Through different experiments, we find that the index can greatly reduce the amount of data that must be retrieved, and will lead to great improvements of performance in large sequence database compared with a sequential search. |
Year | DOI | Venue |
---|---|---|
2005 | 10.1016/j.ins.2004.07.014 | Inf. Sci. |
Keywords | Field | DocType |
dynamic time,stw distance,proposed segment-wise time warping,different experiment,time scaling,lower bound technique,natural time,dynamic programming,corresponding index method,time series data,time warping,indexation,sequential search,spatial index,dynamic time warping,time series,lower bound,similarity search | Time series,Data mining,Similarity measure,Dynamic time warping,Upper and lower bounds,Computer science,Artificial intelligence,Nearest neighbor search,Dynamic programming,Algorithm,Time series database,Linear search,Machine learning | Journal |
Volume | Issue | ISSN |
173 | 1-3 | 0020-0255 |
Citations | PageRank | References |
19 | 0.87 | 21 |
Authors | ||
2 |
Name | Order | Citations | PageRank |
---|---|---|---|
Mi Zhou | 1 | 25 | 1.37 |
Man Hon Wong | 2 | 814 | 233.13 |