Matches in SemOpenAlex for { <https://semopenalex.org/work/W1820176886> ?p ?o ?g. }
- W1820176886 abstract "We study boosting algorithms from a new perspective. We show that the Lagrange dual problems of AdaBoost, LogitBoost and soft-margin LPBoost with generalized hinge loss are all entropy maximization problems. By looking at the dual problems of these boosting algorithms, we show that the success of boosting algorithms can be understood in terms of maintaining a better margin distribution by maximizing margins and at the same time controlling the margin variance.We also theoretically prove that, approximately, AdaBoost maximizes the average margin, instead of the minimum margin. The duality formulation also enables us to develop column generation based optimization algorithms, which are totally corrective. We show that they exhibit almost identical classification results to that of standard stage-wise additive boosting algorithms but with much faster convergence rates. Therefore fewer weak classifiers are needed to build the ensemble using our proposed optimization technique." @default.
- W1820176886 created "2016-06-24" @default.
- W1820176886 creator A5003131173 @default.
- W1820176886 creator A5006294869 @default.
- W1820176886 date "2009-01-23" @default.
- W1820176886 modified "2023-09-27" @default.
- W1820176886 title "A Duality View of Boosting Algorithms" @default.
- W1820176886 cites W1540007258 @default.
- W1820176886 cites W1546961578 @default.
- W1820176886 cites W1570060426 @default.
- W1820176886 cites W1570592793 @default.
- W1820176886 cites W1578080815 @default.
- W1820176886 cites W1918179283 @default.
- W1820176886 cites W1931284523 @default.
- W1820176886 cites W1956110702 @default.
- W1820176886 cites W1970355999 @default.
- W1820176886 cites W1975846642 @default.
- W1820176886 cites W1977848238 @default.
- W1820176886 cites W1982032418 @default.
- W1820176886 cites W1983739894 @default.
- W1820176886 cites W1983874169 @default.
- W1820176886 cites W1988790447 @default.
- W1820176886 cites W2016133707 @default.
- W1820176886 cites W2017170340 @default.
- W1820176886 cites W2024046085 @default.
- W1820176886 cites W2069936846 @default.
- W1820176886 cites W2079484126 @default.
- W1820176886 cites W2098300287 @default.
- W1820176886 cites W2099968818 @default.
- W1820176886 cites W2105180921 @default.
- W1820176886 cites W2122168905 @default.
- W1820176886 cites W2129763352 @default.
- W1820176886 cites W2135125546 @default.
- W1820176886 cites W2143386126 @default.
- W1820176886 cites W2153635508 @default.
- W1820176886 cites W2154462399 @default.
- W1820176886 cites W2157112198 @default.
- W1820176886 cites W2163194396 @default.
- W1820176886 cites W2167277498 @default.
- W1820176886 cites W2168020168 @default.
- W1820176886 cites W2172195373 @default.
- W1820176886 cites W2296319761 @default.
- W1820176886 cites W2751862591 @default.
- W1820176886 cites W298212978 @default.
- W1820176886 cites W3097096317 @default.
- W1820176886 hasPublicationYear "2009" @default.
- W1820176886 type Work @default.
- W1820176886 sameAs 1820176886 @default.
- W1820176886 citedByCount "1" @default.
- W1820176886 countsByYear W18201768862021 @default.
- W1820176886 crossrefType "posted-content" @default.
- W1820176886 hasAuthorship W1820176886A5003131173 @default.
- W1820176886 hasAuthorship W1820176886A5006294869 @default.
- W1820176886 hasConcept C11413529 @default.
- W1820176886 hasConcept C119857082 @default.
- W1820176886 hasConcept C12267149 @default.
- W1820176886 hasConcept C126255220 @default.
- W1820176886 hasConcept C141404830 @default.
- W1820176886 hasConcept C154945302 @default.
- W1820176886 hasConcept C2776330181 @default.
- W1820176886 hasConcept C33923547 @default.
- W1820176886 hasConcept C39891107 @default.
- W1820176886 hasConcept C41008148 @default.
- W1820176886 hasConcept C46686674 @default.
- W1820176886 hasConcept C774472 @default.
- W1820176886 hasConceptScore W1820176886C11413529 @default.
- W1820176886 hasConceptScore W1820176886C119857082 @default.
- W1820176886 hasConceptScore W1820176886C12267149 @default.
- W1820176886 hasConceptScore W1820176886C126255220 @default.
- W1820176886 hasConceptScore W1820176886C141404830 @default.
- W1820176886 hasConceptScore W1820176886C154945302 @default.
- W1820176886 hasConceptScore W1820176886C2776330181 @default.
- W1820176886 hasConceptScore W1820176886C33923547 @default.
- W1820176886 hasConceptScore W1820176886C39891107 @default.
- W1820176886 hasConceptScore W1820176886C41008148 @default.
- W1820176886 hasConceptScore W1820176886C46686674 @default.
- W1820176886 hasConceptScore W1820176886C774472 @default.
- W1820176886 hasLocation W18201768861 @default.
- W1820176886 hasOpenAccess W1820176886 @default.
- W1820176886 hasPrimaryLocation W18201768861 @default.
- W1820176886 hasRelatedWork W1534925209 @default.
- W1820176886 hasRelatedWork W1597529531 @default.
- W1820176886 hasRelatedWork W1603462179 @default.
- W1820176886 hasRelatedWork W1917782903 @default.
- W1820176886 hasRelatedWork W1918179283 @default.
- W1820176886 hasRelatedWork W1974656106 @default.
- W1820176886 hasRelatedWork W2016133707 @default.
- W1820176886 hasRelatedWork W2054697721 @default.
- W1820176886 hasRelatedWork W2098300287 @default.
- W1820176886 hasRelatedWork W2149079208 @default.
- W1820176886 hasRelatedWork W2153656574 @default.
- W1820176886 hasRelatedWork W2154179941 @default.
- W1820176886 hasRelatedWork W2157705457 @default.
- W1820176886 hasRelatedWork W2212091021 @default.
- W1820176886 hasRelatedWork W2340870721 @default.
- W1820176886 hasRelatedWork W2483027298 @default.
- W1820176886 hasRelatedWork W2483585653 @default.
- W1820176886 hasRelatedWork W262212851 @default.
- W1820176886 hasRelatedWork W2952760143 @default.
- W1820176886 hasRelatedWork W3098897816 @default.