Matches in SemOpenAlex for { <https://semopenalex.org/work/W2982070413> ?p ?o ?g. }
- W2982070413 abstract "We consider the classical problem of learning rates for classes with finite VC dimension. It is well known that fast learning rates up to $Oleft(frac{d}{n}right)$ are achievable by the empirical risk minimization algorithm (ERM) if low noise or margin assumptions are satisfied. These usually require the optimal Bayes classifier to be in the class, and it has been shown that when this is not the case, the fast rates cannot be achieved even in the noise free case. In this paper, we further investigate the question of the fast rates under the misspecification, when the Bayes classifier is not in the class (also called the agnostic setting). First, we consider classification with a reject option, namely Chow's reject option model, and show that by slightly lowering the impact of hard instances, a learning rate of order $Oleft(frac{d}{n}log frac{n}{d}right)$ is always achievable in the agnostic setting by a specific learning algorithm. Similar results were only known under special versions of margin assumptions. We also show that the performance of the proposed algorithm is never worse than the performance of ERM. Based on those results, we derive the necessary and sufficient conditions for classification (without a reject option) with fast rates in the agnostic setting achievable by improper learners. This simultaneously extends the work of Massart and N'{e}d'{e}lec (Ann. of Statistics, 2006), which studied this question in the case where the Bayesian optimal rule belongs to the class, and the work of Ben-David and Urner (COLT, 2014), which allows the misspecification but is limited to the no noise setting. Our result also provides the first general setup in statistical learning theory in which an improper learning algorithm may significantly improve the learning rate for non-convex losses." @default.
- W2982070413 created "2019-11-01" @default.
- W2982070413 creator A5014847122 @default.
- W2982070413 creator A5071419221 @default.
- W2982070413 date "2019-10-28" @default.
- W2982070413 modified "2023-09-23" @default.
- W2982070413 title "Fast classification rates without standard margin assumptions" @default.
- W2982070413 cites W1507695921 @default.
- W2982070413 cites W1542886316 @default.
- W2982070413 cites W1575244755 @default.
- W2982070413 cites W1982032418 @default.
- W2982070413 cites W2014902932 @default.
- W2982070413 cites W2023163512 @default.
- W2982070413 cites W2029538739 @default.
- W2982070413 cites W2042024619 @default.
- W2982070413 cites W2043919728 @default.
- W2982070413 cites W2058239479 @default.
- W2982070413 cites W2066688546 @default.
- W2982070413 cites W2095439633 @default.
- W2982070413 cites W2101946573 @default.
- W2982070413 cites W2103619557 @default.
- W2982070413 cites W2104770855 @default.
- W2982070413 cites W2117756453 @default.
- W2982070413 cites W2118377301 @default.
- W2982070413 cites W2126022166 @default.
- W2982070413 cites W2135986977 @default.
- W2982070413 cites W2154637324 @default.
- W2982070413 cites W2154952480 @default.
- W2982070413 cites W2158707174 @default.
- W2982070413 cites W2161813894 @default.
- W2982070413 cites W2603161330 @default.
- W2982070413 cites W2770659572 @default.
- W2982070413 cites W2963007973 @default.
- W2982070413 cites W2963719122 @default.
- W2982070413 cites W2963783614 @default.
- W2982070413 cites W2964282972 @default.
- W2982070413 cites W3011981266 @default.
- W2982070413 cites W3022373602 @default.
- W2982070413 cites W3045059498 @default.
- W2982070413 cites W3046390561 @default.
- W2982070413 cites W3099410262 @default.
- W2982070413 cites W3101940764 @default.
- W2982070413 cites W3105849782 @default.
- W2982070413 cites W358045163 @default.
- W2982070413 cites W568673721 @default.
- W2982070413 cites W593234165 @default.
- W2982070413 hasPublicationYear "2019" @default.
- W2982070413 type Work @default.
- W2982070413 sameAs 2982070413 @default.
- W2982070413 citedByCount "0" @default.
- W2982070413 crossrefType "posted-content" @default.
- W2982070413 hasAuthorship W2982070413A5014847122 @default.
- W2982070413 hasAuthorship W2982070413A5071419221 @default.
- W2982070413 hasConcept C107321475 @default.
- W2982070413 hasConcept C107673813 @default.
- W2982070413 hasConcept C11413529 @default.
- W2982070413 hasConcept C114614502 @default.
- W2982070413 hasConcept C119857082 @default.
- W2982070413 hasConcept C12267149 @default.
- W2982070413 hasConcept C126255220 @default.
- W2982070413 hasConcept C147764199 @default.
- W2982070413 hasConcept C154945302 @default.
- W2982070413 hasConcept C207201462 @default.
- W2982070413 hasConcept C2777212361 @default.
- W2982070413 hasConcept C28826006 @default.
- W2982070413 hasConcept C33676613 @default.
- W2982070413 hasConcept C33923547 @default.
- W2982070413 hasConcept C41008148 @default.
- W2982070413 hasConcept C52001869 @default.
- W2982070413 hasConcept C774472 @default.
- W2982070413 hasConcept C95623464 @default.
- W2982070413 hasConceptScore W2982070413C107321475 @default.
- W2982070413 hasConceptScore W2982070413C107673813 @default.
- W2982070413 hasConceptScore W2982070413C11413529 @default.
- W2982070413 hasConceptScore W2982070413C114614502 @default.
- W2982070413 hasConceptScore W2982070413C119857082 @default.
- W2982070413 hasConceptScore W2982070413C12267149 @default.
- W2982070413 hasConceptScore W2982070413C126255220 @default.
- W2982070413 hasConceptScore W2982070413C147764199 @default.
- W2982070413 hasConceptScore W2982070413C154945302 @default.
- W2982070413 hasConceptScore W2982070413C207201462 @default.
- W2982070413 hasConceptScore W2982070413C2777212361 @default.
- W2982070413 hasConceptScore W2982070413C28826006 @default.
- W2982070413 hasConceptScore W2982070413C33676613 @default.
- W2982070413 hasConceptScore W2982070413C33923547 @default.
- W2982070413 hasConceptScore W2982070413C41008148 @default.
- W2982070413 hasConceptScore W2982070413C52001869 @default.
- W2982070413 hasConceptScore W2982070413C774472 @default.
- W2982070413 hasConceptScore W2982070413C95623464 @default.
- W2982070413 hasOpenAccess W2982070413 @default.
- W2982070413 hasRelatedWork W1522263593 @default.
- W2982070413 hasRelatedWork W2065861851 @default.
- W2982070413 hasRelatedWork W2087950056 @default.
- W2982070413 hasRelatedWork W2104770855 @default.
- W2982070413 hasRelatedWork W2107189314 @default.
- W2982070413 hasRelatedWork W2113399228 @default.
- W2982070413 hasRelatedWork W2150181672 @default.
- W2982070413 hasRelatedWork W2758269203 @default.
- W2982070413 hasRelatedWork W2803642127 @default.
- W2982070413 hasRelatedWork W2950791193 @default.