Supplementary Material Feature Generation for Outlier Detection School of Computing Science Simon Fraser University Vancouver Canada Feature Generation for Outlier Detection aka Propositionalization ID: 1001392
Download Presentation The PPT/PDF document "Tutorial on Learning Bayesian Networks f..." is the property of its rightful owner. Permission is granted to download and print the materials on this web site for personal, non-commercial use only, and to display it on your personal computer provided you do not modify the materials and that you retain all copyright notices contained in the materials. By downloading content from our website, you accept the terms of this agreement.
1. Tutorial on Learning Bayesian Networks for Relational DataSupplementary MaterialFeature Generation for Outlier DetectionSchool of Computing ScienceSimon Fraser UniversityVancouver, Canada
2. Feature Generation for Outlier Detectionaka Propositionalization, Relation EliminationSimilar to feature generation for classificationMain difference: include all first-order random variables, not just the Markov blanket of the class variableRelated work: The Oddball system also extracts a feature matrix from relational information.cite oddball
3. Example: population dataTrue$500KTrue$5MTrue$2MFalsen/a ActsInsalaryFalsen/aFalsen/aFalsen/aFalsen/agender = Mancountry = U.S.gender = Mancountry = U.S.gender = Womancountry = U.S.gender = Womancountry = U.S.runtime = 98 mindrama = trueaction = trueruntime = 111 mindrama = falseaction = true
4. Example: Class Bayesian NetworkPresentation Title At Venuegender(A)ActsIn(A,M)Drama(M)
5. Feature Vectors (I)Presentation Title At Venuegender(A)ActsIn(A,M)Drama(M)Class Bayesian NetworkActsIn(A,M)T01/21/21/2F11/21/21/2Drama(M)T1/21/21/21/2F1/21/21/21/2MovieDramaFargoTKill BillFFeature Matrix for ActsIn(A,M)Feature Matrix for Drama(M)
6. Feature Vectors (II)Presentation Title At Venuegender(A) ActsIn(A,M)Drama(M)MTT001/20MTF0000MFT1/2000MFF1/201/20WTT0000WTF01/201/2WFT01/201/2WFF0000gender(A)ActsIn(A,M)Drama(M)MovieDramaFargoTKill BillF
7. Concatenate all Feature VectorsPresentation Title At Venue01/21/21/211/21/21/21/21/21/21/21/21/21/21/2001/2000001/20001/201/20000001/201/201/201/20000
8. Form Feature MatrixPresentation Title At Venuetranspose to form single-table feature matrix011/21/2001/21/200001/21/21/21/2000001/21/201/21/21/21/21/2001/200001/21/21/21/2000001/21/20