Fast Learning Rates for Plug-In Classifiers

Fast Learning Rates for Plug-In Classifiers

It has been recently shown that, under the margin (or low noise) assumption, there exist classifiers attaining fast rates of convergence of the excess Bayes risk, that is, rates faster than $n^{-1/2}$. The work on this subject has suggested the following two conjectures: (i) the best achievable fast...

Ausführliche Beschreibung

Bibliographische Detailangaben
Veröffentlicht in:The Annals of Statistics. - Institute of Mathematical Statistics. - 35(2007), 2, Seite 608-633
1. Verfasser: Audibert, Jean-Yves (VerfasserIn)
Weitere Verfasser: Tsybakov, Alexandre B.
Format: Online-Aufsatz
Sprache:English
Veröffentlicht: 2007
Zugriff auf das übergeordnete Werk:The Annals of Statistics
Schlagworte:Classification Statistical learning Fast rates of convergence Excess risk Plug-in classifiers Minimax lower bounds Mathematics Behavioral sciences Physical sciences Applied sciences
LEADER 01000caa a22002652 4500
001 JST008965390
003 DE-627
005 20240619180550.0
007 cr uuu---uuuuu
008 150323s2007 xx |||||o 00| ||eng c
035 |a (DE-627)JST008965390 
035 |a (JST)25463570 
040 |a DE-627  |b ger  |c DE-627  |e rakwb 
041 |a eng 
084 |a 62G07  |2 MSC 
084 |a 62G08  |2 MSC 
084 |a 62H05  |2 MSC 
084 |a 68T10  |2 MSC 
100 1 |a Audibert, Jean-Yves  |e verfasserin  |4 aut 
245 1 0 |a Fast Learning Rates for Plug-In Classifiers 
264 1 |c 2007 
336 |a Text  |b txt  |2 rdacontent 
337 |a Computermedien  |b c  |2 rdamedia 
338 |a Online-Ressource  |b cr  |2 rdacarrier 
520 |a It has been recently shown that, under the margin (or low noise) assumption, there exist classifiers attaining fast rates of convergence of the excess Bayes risk, that is, rates faster than $n^{-1/2}$. The work on this subject has suggested the following two conjectures: (i) the best achievable fast rate is of the order n⁻¹, and (ii) the plug-in classifiers generally converge more slowly than the classifiers based on empirical risk minimization. We show that both conjectures are not correct. In particular, we construct plug-in classifiers that can achieve not only fast, but also super-fast rates, that is, rates faster than n⁻¹. We establish minimax lower bounds showing that the obtained rates cannot be improved. 
540 |a Copyright 2007 Institute of Mathematical Statistics 
650 4 |a Classification 
650 4 |a Statistical learning 
650 4 |a Fast rates of convergence 
650 4 |a Excess risk 
650 4 |a Plug-in classifiers 
650 4 |a Minimax lower bounds 
650 4 |a Mathematics  |x Applied mathematics  |x Statistics  |x Applied statistics  |x Inferential statistics  |x Statistical estimation  |x Estimation methods  |x Estimators 
650 4 |a Behavioral sciences  |x Psychology  |x Cognitive psychology  |x Cognitive processes  |x Learning  |x Learning rate 
650 4 |a Mathematics  |x Applied mathematics  |x Statistics  |x Applied statistics  |x Descriptive statistics  |x Statistical distributions  |x Distribution functions  |x Probability distributions 
650 4 |a Mathematics  |x Mathematical expressions  |x Mathematical functions 
650 4 |a Mathematics  |x Pure mathematics  |x Algebra  |x Polynomials 
650 4 |a Mathematics  |x Applied mathematics  |x Game theory  |x Minimax 
650 4 |a Physical sciences  |x Physics  |x Mechanics  |x Density 
650 4 |a Applied sciences  |x Computer science  |x Artificial intelligence  |x Machine learning  |x Perceptron convergence procedure 
650 4 |a Mathematics  |x Pure mathematics  |x Discrete mathematics  |x Number theory  |x Numbers  |x Real numbers  |x Rational numbers  |x Integers 
650 4 |a Mathematics  |x Mathematical analysis  |x Measure theory  |x Lebesgue measures 
650 4 |a Mathematics  |x Applied mathematics  |x Statistics  |x Applied statistics  |x Inferential statistics  |x Statistical estimation  |x Estimation methods  |x Estimators 
650 4 |a Behavioral sciences  |x Psychology  |x Cognitive psychology  |x Cognitive processes  |x Learning  |x Learning rate 
650 4 |a Mathematics  |x Applied mathematics  |x Statistics  |x Applied statistics  |x Descriptive statistics  |x Statistical distributions  |x Distribution functions  |x Probability distributions 
650 4 |a Mathematics  |x Mathematical expressions  |x Mathematical functions 
650 4 |a Mathematics  |x Pure mathematics  |x Algebra  |x Polynomials 
650 4 |a Mathematics  |x Applied mathematics  |x Game theory  |x Minimax 
650 4 |a Physical sciences  |x Physics  |x Mechanics  |x Density 
650 4 |a Applied sciences  |x Computer science  |x Artificial intelligence  |x Machine learning  |x Perceptron convergence procedure 
650 4 |a Mathematics  |x Pure mathematics  |x Discrete mathematics  |x Number theory  |x Numbers  |x Real numbers  |x Rational numbers  |x Integers 
650 4 |a Mathematics  |x Mathematical analysis  |x Measure theory  |x Lebesgue measures  |x Statistical Learning Theory 
655 4 |a research-article 
700 1 |a Tsybakov, Alexandre B.  |e verfasserin  |4 aut 
773 0 8 |i Enthalten in  |t The Annals of Statistics  |d Institute of Mathematical Statistics  |g 35(2007), 2, Seite 608-633  |w (DE-627)270129162  |w (DE-600)1476670-X  |x 00905364  |7 nnns 
773 1 8 |g volume:35  |g year:2007  |g number:2  |g pages:608-633 
856 4 0 |u https://www.jstor.org/stable/25463570  |3 Volltext 
912 |a GBV_USEFLAG_A 
912 |a SYSFLAG_A 
912 |a GBV_JST 
912 |a GBV_ILN_11 
912 |a GBV_ILN_20 
912 |a GBV_ILN_22 
912 |a GBV_ILN_23 
912 |a GBV_ILN_24 
912 |a GBV_ILN_31 
912 |a GBV_ILN_32 
912 |a GBV_ILN_39 
912 |a GBV_ILN_40 
912 |a GBV_ILN_60 
912 |a GBV_ILN_62 
912 |a GBV_ILN_63 
912 |a GBV_ILN_69 
912 |a GBV_ILN_70 
912 |a GBV_ILN_73 
912 |a GBV_ILN_90 
912 |a GBV_ILN_95 
912 |a GBV_ILN_100 
912 |a GBV_ILN_105 
912 |a GBV_ILN_110 
912 |a GBV_ILN_120 
912 |a GBV_ILN_151 
912 |a GBV_ILN_161 
912 |a GBV_ILN_170 
912 |a GBV_ILN_213 
912 |a GBV_ILN_230 
912 |a GBV_ILN_285 
912 |a GBV_ILN_293 
912 |a GBV_ILN_370 
912 |a GBV_ILN_374 
912 |a GBV_ILN_602 
912 |a GBV_ILN_702 
912 |a GBV_ILN_2001 
912 |a GBV_ILN_2003 
912 |a GBV_ILN_2005 
912 |a GBV_ILN_2006 
912 |a GBV_ILN_2007 
912 |a GBV_ILN_2008 
912 |a GBV_ILN_2009 
912 |a GBV_ILN_2010 
912 |a GBV_ILN_2011 
912 |a GBV_ILN_2014 
912 |a GBV_ILN_2015 
912 |a GBV_ILN_2018 
912 |a GBV_ILN_2020 
912 |a GBV_ILN_2021 
912 |a GBV_ILN_2026 
912 |a GBV_ILN_2027 
912 |a GBV_ILN_2044 
912 |a GBV_ILN_2050 
912 |a GBV_ILN_2056 
912 |a GBV_ILN_2057 
912 |a GBV_ILN_2061 
912 |a GBV_ILN_2088 
912 |a GBV_ILN_2107 
912 |a GBV_ILN_2110 
912 |a GBV_ILN_2111 
912 |a GBV_ILN_2190 
912 |a GBV_ILN_2932 
912 |a GBV_ILN_2947 
912 |a GBV_ILN_2949 
912 |a GBV_ILN_2950 
912 |a GBV_ILN_4012 
912 |a GBV_ILN_4035 
912 |a GBV_ILN_4037 
912 |a GBV_ILN_4046 
912 |a GBV_ILN_4112 
912 |a GBV_ILN_4125 
912 |a GBV_ILN_4126 
912 |a GBV_ILN_4242 
912 |a GBV_ILN_4249 
912 |a GBV_ILN_4251 
912 |a GBV_ILN_4305 
912 |a GBV_ILN_4306 
912 |a GBV_ILN_4307 
912 |a GBV_ILN_4313 
912 |a GBV_ILN_4322 
912 |a GBV_ILN_4323 
912 |a GBV_ILN_4324 
912 |a GBV_ILN_4325 
912 |a GBV_ILN_4326 
912 |a GBV_ILN_4335 
912 |a GBV_ILN_4338 
912 |a GBV_ILN_4346 
912 |a GBV_ILN_4367 
912 |a GBV_ILN_4393 
912 |a GBV_ILN_4700 
951 |a AR 
952 |d 35  |j 2007  |e 2  |h 608-633