Rule Extraction from Support Vector Machines: A Sequential Covering Approach Export

@article{citeulike:2459794, abstract = {In this paper, we propose a novel algorithm for rule extraction from support vector machines (SVMs), termed SQRex-SVM. The proposed method extracts rules directly from the support vectors (SVs) of a trained SVM using a modified sequential covering algorithm. Rules are generated based on an ordered search of the most discriminative features, as measured by interclass separation. Rule performance is then evaluated using measured rates of true and false positives and the area under the receiver operating characteristic (ROC) curve (AUC). Results are presented on a number of commonly used data sets that show the rules produced by SQRex-SVM exhibit both improved generalization performance and smaller more comprehensible rule sets compared to both other SVM rule extraction techniques and direct rule learning techniques.}, address = {Piscataway, NJ, USA}, author = {Barakat, Nahla H. and Bradley, Andrew P.}, booktitle = {Knowledge and Data Engineering, IEEE Transactions on}, citeulike-article-id = {2459794}, citeulike-linkout-0 = {http://portal.acm.org/citation.cfm?id=1263361}, citeulike-linkout-1 = {http://dx.doi.org/10.1109/TKDE.2007.190610}, citeulike-linkout-2 = {http://ieeexplore.ieee.org/xpls/abs_all.jsp?arnumber=4161896}, citeulike-linkout-3 = {http://ieeexplore.ieee.org.ezproxy.webfeat.lib.ed.ac.uk/xpls/abs_all.jsp?isnumber=4161893&arnumber=4161896&count=11&index=0}, doi = {10.1109/TKDE.2007.190610}, issn = {1041-4347}, journal = {IEEE Transactions on Knowledge and Data Engineering}, keywords = {phd, rule-extraction, svm}, number = {6}, pages = {729--741}, posted-at = {2009-09-25 11:38:08}, priority = {0}, publisher = {IEEE Educational Activities Department}, title = {Rule Extraction from Support Vector Machines: A Sequential Covering Approach}, url = {http://dx.doi.org/10.1109/TKDE.2007.190610}, volume = {19}, year = {2007} }

TY - JOUR ID - citeulike:2459794 L3 - citeulike-article-id:2459794 N2 - In this paper, we propose a novel algorithm for rule extraction from support vector machines (SVMs), termed SQRex-SVM. The proposed method extracts rules directly from the support vectors (SVs) of a trained SVM using a modified sequential covering algorithm. Rules are generated based on an ordered search of the most discriminative features, as measured by interclass separation. Rule performance is then evaluated using measured rates of true and false positives and the area under the receiver operating characteristic (ROC) curve (AUC). Results are presented on a number of commonly used data sets that show the rules produced by SQRex-SVM exhibit both improved generalization performance and smaller more comprehensible rule sets compared to both other SVM rule extraction techniques and direct rule learning techniques. IS - 6 JF - IEEE Transactions on Knowledge and Data Engineering SN - 1041-4347 AD - Piscataway, NJ, USA EP - 741 TI - Rule Extraction from Support Vector Machines: A Sequential Covering Approach PB - IEEE Educational Activities Department VL - 19 T2 - Knowledge and Data Engineering, IEEE Transactions on SP - 729 KW - phd KW - rule-extraction KW - svm AU - Barakat, Nahla AU - Bradley, Andrew PY - 2007/// UR - http://dx.doi.org/10.1109/TKDE.2007.190610 ER -