?url_ver=Z39.88-2004&rft_val_fmt=info%3Aofi%2Ffmt%3Akev%3Amtx%3Adc&rft.title=Theoretical+analyses+of+cross-validation+error+and+voting+in+instance-based+learning&rft.creator=Turney%2C+Peter+D.&rft.subject=Artificial+Intelligence&rft.subject=Machine+Learning&rft.subject=Statistical+Models&rft.description=This+paper+begins+with+a+general+theory+of+error+in+cross-validation+testing+of+algorithms%0Afor+supervised+learning+from+examples.+It+is+assumed+that+the+examples+are+described+by%0Aattribute-value+pairs%2C+where+the+values+are+symbolic.+Cross-validation+requires+a+set+of%0Atraining+examples+and+a+set+of+testing+examples.+The+value+of+the+attribute+that+is+to+be%0Apredicted+is+known+to+the+learner+in+the+training+set%2C+but+unknown+in+the+testing+set.+The%0Atheory+demonstrates+that+cross-validation+error+has+two+components%3A+error+on+the+training%0Aset+(inaccuracy)+and+sensitivity+to+noise+(instability).%0AThis+general+theory+is+then+applied+to+voting+in+instance-based+learning.+Given+an%0Aexample+in+the+testing+set%2C+a+typical+instance-based+learning+algorithm+predicts+the+designated%0Aattribute+by+voting+among+the+k+nearest+neighbors+(the+k+most+similar+examples)+to%0Athe+testing+example+in+the+training+set.+Voting+is+intended+to+increase+the+stability+(resistance%0Ato+noise)+of+instance-based+learning%2C+but+a+theoretical+analysis+shows+that+there+are%0Acircumstances+in+which+voting+can+be+destabilizing.+The+theory+suggests+ways+to+minimize%0Across-validation+error%2C+by+insuring+that+voting+is+stable+and+does+not+adversely+affect%0Aaccuracy.&rft.date=1994&rft.type=Journal+(Paginated)&rft.type=PeerReviewed&rft.format=application%2Fpdf&rft.identifier=http%3A%2F%2Fcogprints.org%2F1821%2F3%2FNRC-35073.pdf&rft.identifier=++Turney%2C+Peter+D.++(1994)+Theoretical+analyses+of+cross-validation+error+and+voting+in+instance-based+learning.++%5BJournal+(Paginated)%5D+++++&rft.relation=http%3A%2F%2Fcogprints.org%2F1821%2F