Abstract | ||
---|---|---|
Shapelet is a discriminative subsequence of time series. An advanced time series classification method is to integrate shapelet with random forest. However, it shows several limitations. First, random shapelet forest requires a large training cost for split threshold searching. Second, a single shapelet provides limited information for only one branch of the decision tree, resulting in insufficient accuracy and interpretability. Third, randomized ensemble causes interpretability declining. For that, this paper presents Random Pairwise Shapelets Forest (RPSF). RPSF combines a pair of shapelets from different classes to construct random forest. It is more efficient due to omit of threshold search, and more effective due to including of additional information from different classes. Moreover, a discriminability metric, Decomposed Mean Decrease Impurity (DMDI), is proposed to identify influential region for every class. Extensive experiments show that RPSF improves the accuracy and training speed of shapelet forest. Case studies demonstrate the interpretability of our method. |
Year | DOI | Venue |
---|---|---|
2018 | 10.1007/978-3-319-93034-3_6 | ADVANCES IN KNOWLEDGE DISCOVERY AND DATA MINING, PAKDD 2018, PT I |
Keywords | Field | DocType |
Time series classification,Shapelet,Random forest,Interpretability | Pairwise comparison,Decision tree,Interpretability,Computer science,Artificial intelligence,Subsequence,Random forest,Discriminative model,Machine learning,Time series classification | Conference |
Volume | ISSN | Citations |
10937 | 0302-9743 | 0 |
PageRank | References | Authors |
0.34 | 13 | 4 |
Name | Order | Citations | PageRank |
---|---|---|---|
Mohan Shi | 1 | 0 | 0.34 |
Zhihai Wang | 2 | 425 | 28.26 |
Jidong Yuan | 3 | 18 | 6.45 |
Haiyang Liu | 4 | 28 | 10.84 |