Publications
Publication Information
Title | A Generalized Shapelet Learning Method for Unsupervised Multivariate Time Series Clustering |
Authors | Christopher Tennant, Adam Carpenter, Lasitha Vidyaratne, Md. Monibor Rahman, K. Iftekharuddin, Alexander Glandon, Anna Shablina |
JLAB number | JLAB-ACP-21-3323 |
LANL number | (None) |
Other number | DOE/OR/23177-5137 |
Document Type(s) | (Meeting) |
Category: | Computational Physics |
Associated with EIC: | No |
Supported by Jefferson Lab LDRD Funding: | Yes |
LDRD Numbers: | 2020-LDRD-3 |
Funding Source: | Nuclear Physics (NP) |
Meeting Paper compiled for International Joint Conference on Neural Networks | |
Publication Abstract: | Unsupervised multivariate time series clustering is important in many application areas. Among many unsupervised methods Shapelet learning has shown promise for univariate time series signal processing. Discovering suitable Shapelets from a large number of candidate Shapelets has been widely studied for classification of univariate time series signals. However, there is no generalized Shapelet based unsupervised clustering of multivariate time series data. Consequently, this work proposes a generalized Shapelet learning framework for unsupervised multivariate time series clustering. The proposed method utilizes spectral clustering and Shapelet similarity minimization with least square regularization to obtain the optimal multivariate Shapelets for unsupervised clustering. The proposed method is evaluated using an in-house multivariate time series dataset on detection of radio frequency (RF) faults in the Jefferson Labs Continuous Beam Accelerator Facility (CEBAF). The dataset constitutes of three-dimensional time series recordings of three RF fault types. The proposed method shows successful clustering performance with a precision of 0.732 with a standard deviation of 2.3%, recall of 0.7172 with a standard deviation of 1.7%, F-score of 0.732 with a standard deviation of 0.9%, rand index (RI) score of 0.812 with standard deviation of 0.9%, and an average normalize mutual information (NMI) of 0.56 with a standard deviation of 3.6%, in a five-fold cross validation evaluation. |
Experiment Numbers: | other |
Group: | Ctr for Adv Stud of Accel |
Document: | |
DOI: | |
Accepted Manuscript: | |
Supporting Documents: | |
Supporting Datasets: |