Time Series Cluster Kernel for Learning Similarities between Multivariate Time Series with Missing Data. Øyvind Mikalsen, K., Maria Bianchi, F., Soguero-Ruiz, C., & Jenssen, R. Technical Report Paper abstract bibtex Similarity-based approaches represent a promising direction for time series analysis. However , many such methods rely on parameter tuning, and some have shortcomings if the time series are multivariate (MTS), due to dependencies between attributes, or the time series contain missing data. In this paper, we address these challenges within the powerful context of kernel methods by proposing the robust time series cluster kernel (TCK). The approach taken leverages the missing data handling properties of Gaussian mixture models (GMM) augmented with informative prior distributions. An ensemble learning approach is exploited to ensure robustness to parameters by combining the clustering results of many GMM to form the final kernel. We evaluate the TCK on synthetic and real data and compare to other state-of-the-art techniques. The experimental results demonstrate that the TCK is robust to parameter choices, provides competitive results for MTS without missing data and outstanding results for missing data.
@techreport{
title = {Time Series Cluster Kernel for Learning Similarities between Multivariate Time Series with Missing Data},
type = {techreport},
identifiers = {[object Object]},
keywords = {Ensemble learning,Gaussian mixture models,Kernel methods,Missing data,Multivariate time series,Similarity measures},
id = {441115b6-f746-3d82-b1fc-5ccf6bf74b96},
created = {2018-11-19T15:49:15.135Z},
file_attached = {true},
profile_id = {00330907-7fcb-39a3-a388-db79231d75d8},
group_id = {a4d8c5ee-7397-327b-a54b-330e0b4d7061},
last_modified = {2018-11-19T15:49:15.245Z},
read = {false},
starred = {false},
authored = {false},
confirmed = {false},
hidden = {false},
private_publication = {false},
abstract = {Similarity-based approaches represent a promising direction for time series analysis. However , many such methods rely on parameter tuning, and some have shortcomings if the time series are multivariate (MTS), due to dependencies between attributes, or the time series contain missing data. In this paper, we address these challenges within the powerful context of kernel methods by proposing the robust time series cluster kernel (TCK). The approach taken leverages the missing data handling properties of Gaussian mixture models (GMM) augmented with informative prior distributions. An ensemble learning approach is exploited to ensure robustness to parameters by combining the clustering results of many GMM to form the final kernel. We evaluate the TCK on synthetic and real data and compare to other state-of-the-art techniques. The experimental results demonstrate that the TCK is robust to parameter choices, provides competitive results for MTS without missing data and outstanding results for missing data.},
bibtype = {techreport},
author = {Øyvind Mikalsen, Karl and Maria Bianchi, Filippo and Soguero-Ruiz, Cristina and Jenssen, Robert}
}
Downloads: 0
{"_id":"dZQze8aNiukozMYFZ","bibbaseid":"yvindmikalsen-mariabianchi-sogueroruiz-jenssen-timeseriesclusterkernelforlearningsimilaritiesbetweenmultivariatetimeserieswithmissingdata","downloads":0,"creationDate":"2018-11-19T15:49:29.290Z","title":"Time Series Cluster Kernel for Learning Similarities between Multivariate Time Series with Missing Data","author_short":["Øyvind Mikalsen, K.","Maria Bianchi, F.","Soguero-Ruiz, C.","Jenssen, R."],"year":null,"bibtype":"techreport","biburl":null,"bibdata":{"title":"Time Series Cluster Kernel for Learning Similarities between Multivariate Time Series with Missing Data","type":"techreport","identifiers":"[object Object]","keywords":"Ensemble learning,Gaussian mixture models,Kernel methods,Missing data,Multivariate time series,Similarity measures","id":"441115b6-f746-3d82-b1fc-5ccf6bf74b96","created":"2018-11-19T15:49:15.135Z","file_attached":"true","profile_id":"00330907-7fcb-39a3-a388-db79231d75d8","group_id":"a4d8c5ee-7397-327b-a54b-330e0b4d7061","last_modified":"2018-11-19T15:49:15.245Z","read":false,"starred":false,"authored":false,"confirmed":false,"hidden":false,"private_publication":false,"abstract":"Similarity-based approaches represent a promising direction for time series analysis. However , many such methods rely on parameter tuning, and some have shortcomings if the time series are multivariate (MTS), due to dependencies between attributes, or the time series contain missing data. In this paper, we address these challenges within the powerful context of kernel methods by proposing the robust time series cluster kernel (TCK). The approach taken leverages the missing data handling properties of Gaussian mixture models (GMM) augmented with informative prior distributions. An ensemble learning approach is exploited to ensure robustness to parameters by combining the clustering results of many GMM to form the final kernel. We evaluate the TCK on synthetic and real data and compare to other state-of-the-art techniques. The experimental results demonstrate that the TCK is robust to parameter choices, provides competitive results for MTS without missing data and outstanding results for missing data.","bibtype":"techreport","author":"Øyvind Mikalsen, Karl and Maria Bianchi, Filippo and Soguero-Ruiz, Cristina and Jenssen, Robert","bibtex":"@techreport{\n title = {Time Series Cluster Kernel for Learning Similarities between Multivariate Time Series with Missing Data},\n type = {techreport},\n identifiers = {[object Object]},\n keywords = {Ensemble learning,Gaussian mixture models,Kernel methods,Missing data,Multivariate time series,Similarity measures},\n id = {441115b6-f746-3d82-b1fc-5ccf6bf74b96},\n created = {2018-11-19T15:49:15.135Z},\n file_attached = {true},\n profile_id = {00330907-7fcb-39a3-a388-db79231d75d8},\n group_id = {a4d8c5ee-7397-327b-a54b-330e0b4d7061},\n last_modified = {2018-11-19T15:49:15.245Z},\n read = {false},\n starred = {false},\n authored = {false},\n confirmed = {false},\n hidden = {false},\n private_publication = {false},\n abstract = {Similarity-based approaches represent a promising direction for time series analysis. However , many such methods rely on parameter tuning, and some have shortcomings if the time series are multivariate (MTS), due to dependencies between attributes, or the time series contain missing data. In this paper, we address these challenges within the powerful context of kernel methods by proposing the robust time series cluster kernel (TCK). The approach taken leverages the missing data handling properties of Gaussian mixture models (GMM) augmented with informative prior distributions. An ensemble learning approach is exploited to ensure robustness to parameters by combining the clustering results of many GMM to form the final kernel. We evaluate the TCK on synthetic and real data and compare to other state-of-the-art techniques. The experimental results demonstrate that the TCK is robust to parameter choices, provides competitive results for MTS without missing data and outstanding results for missing data.},\n bibtype = {techreport},\n author = {Øyvind Mikalsen, Karl and Maria Bianchi, Filippo and Soguero-Ruiz, Cristina and Jenssen, Robert}\n}","author_short":["Øyvind Mikalsen, K.","Maria Bianchi, F.","Soguero-Ruiz, C.","Jenssen, R."],"urls":{"Paper":"https://bibbase.org/service/mendeley/00330907-7fcb-39a3-a388-db79231d75d8/file/0bf0b2e1-bb41-33c9-6724-e90546ae0f0a/Time_Series_Cluster_Kernel_for_Learning_Similarities_between_Multivariate_Time_Series_with_Missing_Data.pdf.pdf"},"bibbaseid":"yvindmikalsen-mariabianchi-sogueroruiz-jenssen-timeseriesclusterkernelforlearningsimilaritiesbetweenmultivariatetimeserieswithmissingdata","role":"author","keyword":["Ensemble learning","Gaussian mixture models","Kernel methods","Missing data","Multivariate time series","Similarity measures"],"downloads":0},"search_terms":["time","series","cluster","kernel","learning","similarities","between","multivariate","time","series","missing","data","øyvind mikalsen","maria bianchi","soguero-ruiz","jenssen"],"keywords":["ensemble learning","gaussian mixture models","kernel methods","missing data","multivariate time series","similarity measures"],"authorIDs":[]}