{"@type": "dcat:Dataset", "accessLevel": "public", "accrualPeriodicity": "irregular", "bureauCode": ["026:00"], "contactPoint": {"@type": "vcard:Contact", "fn": "Santanu Das", "hasEmail": "mailto:Santanu.Das-1@nasa.gov"}, "description": "One-class nu-Support Vector machine (SVMs) learning technique maps the \r\ninput data into a much higher dimensional space and then uses a small \r\nportion of the training data (support vectors) to parametrize the \r\ndecision surface that can linearly separate nu fraction of training \r\npoints (labeled as anomalies) from the rest. The exact solution of \r\nstandard one-class nu SVMs assigns (at least) nu fraction of training \r\npoints as support vectors. However some of these support vectors may be \r\nunnecessary or redundant. Hence the computational issue turns alarming \r\nespecially when SVMs based novelty detectors with nonlinear kernels are \r\ntrained on data sets of huge size. The proposed nu-Anomica algorithm can \r\nsolve this problem. The idea is to train the machine such that it can \r\nprovide a close approximation to the exact decision plane using far less \r\nnumber of training points and without loosing much of the generalization \r\nperformance of the classical approach. The developed procedure closely \r\npreserves the accuracy of standard One-class nu-SVMs while reducing both \r\ntraining time and test time by several factors.", "distribution": [{"@type": "dcat:Distribution", "description": "\u03bd-Anomica: A Fast Support Vector based Novelty Detection Technique", "downloadURL": "https://c3.nasa.gov/dashlink/static/media/algorithm/PID1020575.pdf", "format": "PDF", "mediaType": "application/pdf", "title": "PID1020575.pdf"}, {"@type": "dcat:Distribution", "description": "ICDM 2009, presentation", "downloadURL": "https://c3.nasa.gov/dashlink/static/media/algorithm/ICDM_nuAnomica_ver5.pdf", "format": "PDF", "mediaType": "application/pdf", "title": "ICDM_nuAnomica_ver5.pdf"}, {"@type": "dcat:Distribution", "description": "nu-Anomica code (Matlab)", "downloadURL": "https://c3.nasa.gov/dashlink/static/media/algorithm/nu-anomica.zip", "format": "application/x-zip-compressed", "mediaType": "application/x-zip-compressed", "title": "nu-anomica.zip"}], "identifier": "DASHLINK_131", "issued": "2010-09-10", "keyword": ["ames", "dashlink", "nasa"], "landingPage": "https://c3.nasa.gov/dashlink/resources/131/", "modified": "2025-04-01", "programCode": ["026:029"], "publisher": {"@type": "org:Organization", "name": "Dashlink"}, "title": "nu-Anomica algorithm"}