Active learning is widely used to select which examples from a pool should be ... We present the BaBiES criterion, an approximation of Bayesian A-optimal ...
University of Pennsylvania
ScholarlyCommons Technical Reports (CIS)
Department of Computer & Information Science
January 2004
Bayesian Example Selection Using BaBiES Andrew I. Schein University of Pennsylvania
S. Ted Sandler University of Pennsylvania
Lyle H. Ungar University of Pennsylvania, ungar@cis.upenn.edu
Follow this and additional works at: http://repository.upenn.edu/cis_reports Recommended Citation Andrew I. Schein, S. Ted Sandler, and Lyle H. Ungar, "Bayesian Example Selection Using BaBiES", . January 2004.
University of Pennsylvania Department of Computer and Information Science Technical Report No. MS-CIS-04-08. This paper is posted at ScholarlyCommons. http://repository.upenn.edu/cis_reports/10 For more information, please contact libraryrepository@pobox.upenn.edu.
Bayesian Example Selection Using BaBiES Abstract
Active learning is widely used to select which examples from a pool should be labeled to give best results when learning predictive models. It is, however, sometimes desirable to choose examples before any labeling or machine learning has occurred. The optimal experimental design literature has many theoretically attractive optimality criteria for example selection, but most are intractable when working with large numbers of predictive features. We present the BaBiES criterion, an approximation of Bayesian A-optimal design for linear regression using binary predictors, which is both simple and extremely fast. Empirical evaluations demonstrate that, in spite of selecting all examples prior to learning, BaBiES is competitive with standard active learning methods for a variety of document classification tasks. Comments
University of Pennsylvania Department of Computer and Information Science Technical Report No. MSCIS-04-08.
This technical report is available at ScholarlyCommons: http://repository.upenn.edu/cis_reports/10
Béê ë Að{ë¢+ë [ô û î ð A ù B®öð{îúò{êþ gGmnXd¯dad`YX[}õA} Xdx\ÀZ]a1±'Xd`IgÇ}nXx{\pad}neG^_`:¸ sutGUµXdWGWGZfaad}p)aeG`3\fx Xd`Igl\fZfXdmq`Imq`G¯xU\8bkU®X1\]eIZ]U°p)aeG`3\fx gSU xfp)Zfmn´Gmq`I¯\ftGU_`3eI^´RUZ'a[bA\]mn^_Urx'XWGZ]U®gSm p\]a!Z aSprp)eGZ¢xÍmq` \ftGUÀWRa>ad}AXd`Ig»\]Z¢X[mn`Gmn`G¯NxU\ZfUrx]WYU®p\]mnwdUr}qz¸sutGUZ]U®xeG}q\]mn`G¯ a´ ] Urp)\]mnwdUÍbkeG`Ip)\]mnad`i\]a_^_mn`Gmq^_Um eU m x f ÅuXdÅm ÆAo !0y~)È Î #s"À !{¡ ( ) $%'( &