A unifying framework for detecting outliers and change points from non-stationary time series data. Yamanishi, K. & Takeuchi, J. In Proceedings of the eighth ACM SIGKDD international conference on Knowledge discovery and data mining, of KDD '02, pages 676–681, New York, NY, USA, July, 2002. Association for Computing Machinery. ![A unifying framework for detecting outliers and change points from non-stationary time series data [link]](https://bibbase.org/img/filetypes/link.svg "link")
Paper doi abstract bibtex We are concerned with the issues of outlier detection and change point detection from a data stream. In the area of data mining, there have been increased interest in these issues since the former is related to fraud detection, rare event discovery, etc., while the latter is related to event/trend by change detection, activity monitoring, etc. Specifically, it is important to consider the situation where the data source is non-stationary, since the nature of data source may change over time in real applications. Although in most previous work outlier detection and change point detection have not been related explicitly, this paper presents a unifying framework for dealing with both of them on the basis of the theory of on-line learning of non-stationary time series. In this framework a probabilistic model of the data source is incrementally learned using an on-line discounting learning algorithm, which can track the changing data source adaptively by forgetting the effect of past data gradually. Then the score for any given data is calculated to measure its deviation from the learned model, with a higher score indicating a high possibility of being an outlier. Further change points in a data stream are detected by applying this scoring method into a time series of moving averaged losses for prediction using the learned model. Specifically we develop an efficient algorithms for on-line discounting learning of auto-regression models from time series data, and demonstrate the validity of our framework through simulation and experimental applications to stock market data analysis.
@inproceedings{yamanishi_unifying_2002,
address = {New York, NY, USA},
series = {{KDD} '02},
title = {A unifying framework for detecting outliers and change points from non-stationary time series data},
isbn = {978-1-58113-567-1},
url = {https://doi.org/10.1145/775047.775148},
doi = {10.1145/775047.775148},
abstract = {We are concerned with the issues of outlier detection and change point detection from a data stream. In the area of data mining, there have been increased interest in these issues since the former is related to fraud detection, rare event discovery, etc., while the latter is related to event/trend by change detection, activity monitoring, etc. Specifically, it is important to consider the situation where the data source is non-stationary, since the nature of data source may change over time in real applications. Although in most previous work outlier detection and change point detection have not been related explicitly, this paper presents a unifying framework for dealing with both of them on the basis of the theory of on-line learning of non-stationary time series. In this framework a probabilistic model of the data source is incrementally learned using an on-line discounting learning algorithm, which can track the changing data source adaptively by forgetting the effect of past data gradually. Then the score for any given data is calculated to measure its deviation from the learned model, with a higher score indicating a high possibility of being an outlier. Further change points in a data stream are detected by applying this scoring method into a time series of moving averaged losses for prediction using the learned model. Specifically we develop an efficient algorithms for on-line discounting learning of auto-regression models from time series data, and demonstrate the validity of our framework through simulation and experimental applications to stock market data analysis.},
urldate = {2020-10-04},
booktitle = {Proceedings of the eighth {ACM} {SIGKDD} international conference on {Knowledge} discovery and data mining},
publisher = {Association for Computing Machinery},
author = {Yamanishi, Kenji and Takeuchi, Jun-ichi},
month = jul,
year = {2002},
pages = {676--681},
}
Downloads: 0
{"_id":"Y2gCWH26qusXY7qJr","bibbaseid":"yamanishi-takeuchi-aunifyingframeworkfordetectingoutliersandchangepointsfromnonstationarytimeseriesdata-2002","author_short":["Yamanishi, K.","Takeuchi, J."],"bibdata":{"bibtype":"inproceedings","type":"inproceedings","address":"New York, NY, USA","series":"KDD '02","title":"A unifying framework for detecting outliers and change points from non-stationary time series data","isbn":"978-1-58113-567-1","url":"https://doi.org/10.1145/775047.775148","doi":"10.1145/775047.775148","abstract":"We are concerned with the issues of outlier detection and change point detection from a data stream. In the area of data mining, there have been increased interest in these issues since the former is related to fraud detection, rare event discovery, etc., while the latter is related to event/trend by change detection, activity monitoring, etc. Specifically, it is important to consider the situation where the data source is non-stationary, since the nature of data source may change over time in real applications. Although in most previous work outlier detection and change point detection have not been related explicitly, this paper presents a unifying framework for dealing with both of them on the basis of the theory of on-line learning of non-stationary time series. In this framework a probabilistic model of the data source is incrementally learned using an on-line discounting learning algorithm, which can track the changing data source adaptively by forgetting the effect of past data gradually. Then the score for any given data is calculated to measure its deviation from the learned model, with a higher score indicating a high possibility of being an outlier. Further change points in a data stream are detected by applying this scoring method into a time series of moving averaged losses for prediction using the learned model. Specifically we develop an efficient algorithms for on-line discounting learning of auto-regression models from time series data, and demonstrate the validity of our framework through simulation and experimental applications to stock market data analysis.","urldate":"2020-10-04","booktitle":"Proceedings of the eighth ACM SIGKDD international conference on Knowledge discovery and data mining","publisher":"Association for Computing Machinery","author":[{"propositions":[],"lastnames":["Yamanishi"],"firstnames":["Kenji"],"suffixes":[]},{"propositions":[],"lastnames":["Takeuchi"],"firstnames":["Jun-ichi"],"suffixes":[]}],"month":"July","year":"2002","pages":"676–681","bibtex":"@inproceedings{yamanishi_unifying_2002,\n\taddress = {New York, NY, USA},\n\tseries = {{KDD} '02},\n\ttitle = {A unifying framework for detecting outliers and change points from non-stationary time series data},\n\tisbn = {978-1-58113-567-1},\n\turl = {https://doi.org/10.1145/775047.775148},\n\tdoi = {10.1145/775047.775148},\n\tabstract = {We are concerned with the issues of outlier detection and change point detection from a data stream. In the area of data mining, there have been increased interest in these issues since the former is related to fraud detection, rare event discovery, etc., while the latter is related to event/trend by change detection, activity monitoring, etc. Specifically, it is important to consider the situation where the data source is non-stationary, since the nature of data source may change over time in real applications. Although in most previous work outlier detection and change point detection have not been related explicitly, this paper presents a unifying framework for dealing with both of them on the basis of the theory of on-line learning of non-stationary time series. In this framework a probabilistic model of the data source is incrementally learned using an on-line discounting learning algorithm, which can track the changing data source adaptively by forgetting the effect of past data gradually. Then the score for any given data is calculated to measure its deviation from the learned model, with a higher score indicating a high possibility of being an outlier. Further change points in a data stream are detected by applying this scoring method into a time series of moving averaged losses for prediction using the learned model. Specifically we develop an efficient algorithms for on-line discounting learning of auto-regression models from time series data, and demonstrate the validity of our framework through simulation and experimental applications to stock market data analysis.},\n\turldate = {2020-10-04},\n\tbooktitle = {Proceedings of the eighth {ACM} {SIGKDD} international conference on {Knowledge} discovery and data mining},\n\tpublisher = {Association for Computing Machinery},\n\tauthor = {Yamanishi, Kenji and Takeuchi, Jun-ichi},\n\tmonth = jul,\n\tyear = {2002},\n\tpages = {676--681},\n}\n\n","author_short":["Yamanishi, K.","Takeuchi, J."],"key":"yamanishi_unifying_2002","id":"yamanishi_unifying_2002","bibbaseid":"yamanishi-takeuchi-aunifyingframeworkfordetectingoutliersandchangepointsfromnonstationarytimeseriesdata-2002","role":"author","urls":{"Paper":"https://doi.org/10.1145/775047.775148"},"metadata":{"authorlinks":{}}},"bibtype":"inproceedings","biburl":"https://bibbase.org/zotero/mh_lenguyen","dataSources":["iwKepCrWBps7ojhDx"],"keywords":[],"search_terms":["unifying","framework","detecting","outliers","change","points","non","stationary","time","series","data","yamanishi","takeuchi"],"title":"A unifying framework for detecting outliers and change points from non-stationary time series data","year":2002}