Adaptive Anomaly Detection with Kernel Eigenspace Splitting and Merging. O'Reilly, C., Gluhak, A., & Imran, M. IEEE Transactions on Knowledge and Data Engineering, 27(1):3--16, January, 2015. 00000doi abstract bibtex Kernel principal component analysis and the reconstruction error is an effective anomaly detection technique for non-linear data sets. In an environment where a phenomenon is generating data that is non-stationary, anomaly detection requires a recomputation of the kernel eigenspace in order to represent the current data distribution. Recomputation is a computationally complex operation and reducing computational complexity is therefore a key challenge. In this paper, we propose an algorithm that is able to accurately remove data from a kernel eigenspace without performing a batch recomputation. Coupled with a kernel eigenspace update, we demonstrate that our technique is able to remove and add data to a kernel eigenspace more accurately than existing techniques. An adaptive version determines an appropriately sized sliding window of data and when a model update is necessary. Experimental evaluations on both synthetic and real-world data sets demonstrate the superior performance of the proposed approach in comparison to alternative incremental KPCA approaches and alternative anomaly detection techniques.
@article{ oreilly_adaptive_2015,
title = {Adaptive {Anomaly} {Detection} with {Kernel} {Eigenspace} {Splitting} and {Merging}},
volume = {27},
issn = {1041-4347},
doi = {10.1109/TKDE.2014.2324594},
abstract = {Kernel principal component analysis and the reconstruction error is an effective anomaly detection technique for non-linear data sets. In an environment where a phenomenon is generating data that is non-stationary, anomaly detection requires a recomputation of the kernel eigenspace in order to represent the current data distribution. Recomputation is a computationally complex operation and reducing computational complexity is therefore a key challenge. In this paper, we propose an algorithm that is able to accurately remove data from a kernel eigenspace without performing a batch recomputation. Coupled with a kernel eigenspace update, we demonstrate that our technique is able to remove and add data to a kernel eigenspace more accurately than existing techniques. An adaptive version determines an appropriately sized sliding window of data and when a model update is necessary. Experimental evaluations on both synthetic and real-world data sets demonstrate the superior performance of the proposed approach in comparison to alternative incremental KPCA approaches and alternative anomaly detection techniques.},
number = {1},
journal = {IEEE Transactions on Knowledge and Data Engineering},
author = {O'Reilly, C. and Gluhak, A. and Imran, M.A.},
month = {January},
year = {2015},
note = {00000},
pages = {3--16}
}
Downloads: 0
{"_id":"9DpkcomWAqsRr7jx8","bibbaseid":"oreilly-gluhak-imran-adaptiveanomalydetectionwithkerneleigenspacesplittingandmerging-2015","downloads":0,"creationDate":"2015-09-03T07:18:46.971Z","title":"Adaptive Anomaly Detection with Kernel Eigenspace Splitting and Merging","author_short":["O'Reilly, C.","Gluhak, A.","Imran, M."],"year":2015,"bibtype":"article","biburl":"http://bibbase.org/zotero/fred.qi","bibdata":{"abstract":"Kernel principal component analysis and the reconstruction error is an effective anomaly detection technique for non-linear data sets. In an environment where a phenomenon is generating data that is non-stationary, anomaly detection requires a recomputation of the kernel eigenspace in order to represent the current data distribution. Recomputation is a computationally complex operation and reducing computational complexity is therefore a key challenge. In this paper, we propose an algorithm that is able to accurately remove data from a kernel eigenspace without performing a batch recomputation. Coupled with a kernel eigenspace update, we demonstrate that our technique is able to remove and add data to a kernel eigenspace more accurately than existing techniques. An adaptive version determines an appropriately sized sliding window of data and when a model update is necessary. Experimental evaluations on both synthetic and real-world data sets demonstrate the superior performance of the proposed approach in comparison to alternative incremental KPCA approaches and alternative anomaly detection techniques.","author":["O'Reilly, C.","Gluhak, A.","Imran, M.A."],"author_short":["O'Reilly, C.","Gluhak, A.","Imran, M."],"bibtex":"@article{ oreilly_adaptive_2015,\n title = {Adaptive {Anomaly} {Detection} with {Kernel} {Eigenspace} {Splitting} and {Merging}},\n volume = {27},\n issn = {1041-4347},\n doi = {10.1109/TKDE.2014.2324594},\n abstract = {Kernel principal component analysis and the reconstruction error is an effective anomaly detection technique for non-linear data sets. In an environment where a phenomenon is generating data that is non-stationary, anomaly detection requires a recomputation of the kernel eigenspace in order to represent the current data distribution. Recomputation is a computationally complex operation and reducing computational complexity is therefore a key challenge. In this paper, we propose an algorithm that is able to accurately remove data from a kernel eigenspace without performing a batch recomputation. Coupled with a kernel eigenspace update, we demonstrate that our technique is able to remove and add data to a kernel eigenspace more accurately than existing techniques. An adaptive version determines an appropriately sized sliding window of data and when a model update is necessary. Experimental evaluations on both synthetic and real-world data sets demonstrate the superior performance of the proposed approach in comparison to alternative incremental KPCA approaches and alternative anomaly detection techniques.},\n number = {1},\n journal = {IEEE Transactions on Knowledge and Data Engineering},\n author = {O'Reilly, C. and Gluhak, A. and Imran, M.A.},\n month = {January},\n year = {2015},\n note = {00000},\n pages = {3--16}\n}","bibtype":"article","doi":"10.1109/TKDE.2014.2324594","id":"oreilly_adaptive_2015","issn":"1041-4347","journal":"IEEE Transactions on Knowledge and Data Engineering","key":"oreilly_adaptive_2015","month":"January","note":"00000","number":"1","pages":"3--16","title":"Adaptive Anomaly Detection with Kernel Eigenspace Splitting and Merging","type":"article","volume":"27","year":"2015","bibbaseid":"oreilly-gluhak-imran-adaptiveanomalydetectionwithkerneleigenspacesplittingandmerging-2015","role":"author","urls":{},"downloads":0},"search_terms":["adaptive","anomaly","detection","kernel","eigenspace","splitting","merging","o'reilly","gluhak","imran"],"keywords":[],"authorIDs":[],"dataSources":["y8en6y5RHukfeLuPH"]}