Optimal convergence rates of deep neural networks in a classification setting
We establish optimal convergence rates up to a log factor for a class of deep neural networks in a classification setting under a restraint sometimes referred to as the Tsybakov noise condition. We construct classifiers based on empirical risk minimization in a general setting where the boundary of...
Gespeichert in:
| 1. Verfasser: | |
|---|---|
| Dokumenttyp: | Article (Journal) |
| Sprache: | Englisch |
| Veröffentlicht: |
7 December 2023
|
| In: |
Electronic journal of statistics
Year: 2023, Jahrgang: 17, Heft: 2, Pages: 3613-3659 |
| ISSN: | 1935-7524 |
| DOI: | 10.1214/23-EJS2187 |
| Online-Zugang: | Verlag, kostenfrei, Volltext: https://doi.org/10.1214/23-EJS2187 Verlag, kostenfrei, Volltext: https://projecteuclid.org/journals/electronic-journal-of-statistics/volume-17/issue-2/Optimal-convergence-rates-of-deep-neural-networks-in-a-classification/10.1214/23-EJS2187.full 
|
| Verfasserangaben: | Joseph T. Meyer |
MARC
| LEADER | 00000caa a2200000 c 4500 | ||
|---|---|---|---|
| 001 | 1885863268 | ||
| 003 | DE-627 | ||
| 005 | 20240703164722.0 | ||
| 007 | cr uuu---uuuuu | ||
| 008 | 240415s2023 xx |||||o 00| ||eng c | ||
| 024 | 7 | |a 10.1214/23-EJS2187 |2 doi | |
| 035 | |a (DE-627)1885863268 | ||
| 035 | |a (DE-599)KXP1885863268 | ||
| 035 | |a (OCoLC)1443668941 | ||
| 040 | |a DE-627 |b ger |c DE-627 |e rda | ||
| 041 | |a eng | ||
| 084 | |a 27 |2 sdnb | ||
| 100 | 1 | |a Meyer, Joseph Theo |d 1993- |e VerfasserIn |0 (DE-588)1305413032 |0 (DE-627)1860989233 |4 aut | |
| 245 | 1 | 0 | |a Optimal convergence rates of deep neural networks in a classification setting |c Joseph T. Meyer |
| 264 | 1 | |c 7 December 2023 | |
| 300 | |a 47 | ||
| 336 | |a Text |b txt |2 rdacontent | ||
| 337 | |a Computermedien |b c |2 rdamedia | ||
| 338 | |a Online-Ressource |b cr |2 rdacarrier | ||
| 500 | |a Gesehen am 15.04.2024 | ||
| 520 | |a We establish optimal convergence rates up to a log factor for a class of deep neural networks in a classification setting under a restraint sometimes referred to as the Tsybakov noise condition. We construct classifiers based on empirical risk minimization in a general setting where the boundary of the Bayes rule can be approximated well by neural networks. Corresponding rates of convergence are proven with respect to the misclassification error using an additional condition that acts as a requirement for the “correct noise exponent”. It is then shown that these rates are optimal in the minimax sense. For other estimation procedures, similar convergence rates have been established. Our first main contribution is to prove that the rates are optimal under the additional condition. Secondly, our main theorem establishes almost optimal rates in a generalized setting. We use this to show optimal rates which circumvent the curse of dimensionality. | ||
| 650 | 4 | |a 62C20 | |
| 650 | 4 | |a 62G05 | |
| 650 | 4 | |a classification | |
| 650 | 4 | |a Deep neural networks | |
| 650 | 4 | |a Tsybakov noise condition | |
| 773 | 0 | 8 | |i Enthalten in |t Electronic journal of statistics |d Ithaca, NY : Cornell University Library, 2007 |g 17(2023), 2, Seite 3613-3659 |h Online-Ressource |w (DE-627)538998830 |w (DE-600)2381001-4 |w (DE-576)28134714X |x 1935-7524 |7 nnas |a Optimal convergence rates of deep neural networks in a classification setting |
| 773 | 1 | 8 | |g volume:17 |g year:2023 |g number:2 |g pages:3613-3659 |g extent:47 |a Optimal convergence rates of deep neural networks in a classification setting |
| 856 | 4 | 0 | |u https://doi.org/10.1214/23-EJS2187 |x Verlag |x Resolving-System |z kostenfrei |3 Volltext |
| 856 | 4 | 0 | |u https://projecteuclid.org/journals/electronic-journal-of-statistics/volume-17/issue-2/Optimal-convergence-rates-of-deep-neural-networks-in-a-classification/10.1214/23-EJS2187.full |x Verlag |z kostenfrei |3 Volltext |
| 951 | |a AR | ||
| 992 | |a 20240415 | ||
| 993 | |a Article | ||
| 994 | |a 2023 | ||
| 998 | |g 1305413032 |a Meyer, Joseph Theo |m 1305413032:Meyer, Joseph Theo |d 110000 |d 110400 |e 110000PM1305413032 |e 110400PM1305413032 |k 0/110000/ |k 1/110000/110400/ |p 1 |x j |y j | ||
| 999 | |a KXP-PPN1885863268 |e 4512232532 | ||
| BIB | |a Y | ||
| SER | |a journal | ||
| JSO | |a {"recId":"1885863268","language":["eng"],"type":{"media":"Online-Ressource","bibl":"article-journal"},"note":["Gesehen am 15.04.2024"],"person":[{"family":"Meyer","given":"Joseph Theo","display":"Meyer, Joseph Theo","roleDisplay":"VerfasserIn","role":"aut"}],"title":[{"title_sort":"Optimal convergence rates of deep neural networks in a classification setting","title":"Optimal convergence rates of deep neural networks in a classification setting"}],"relHost":[{"origin":[{"publisherPlace":"Ithaca, NY","dateIssuedDisp":"2007-","publisher":"Cornell University Library","dateIssuedKey":"2007"}],"id":{"issn":["1935-7524"],"eki":["538998830"],"zdb":["2381001-4"]},"physDesc":[{"extent":"Online-Ressource"}],"title":[{"subtitle":"EJS","title":"Electronic journal of statistics","title_sort":"Electronic journal of statistics"}],"pubHistory":["1.2007 -"],"part":{"volume":"17","text":"17(2023), 2, Seite 3613-3659","extent":"47","year":"2023","issue":"2","pages":"3613-3659"},"titleAlt":[{"title":"EJS"}],"type":{"media":"Online-Ressource","bibl":"periodical"},"disp":"Optimal convergence rates of deep neural networks in a classification settingElectronic journal of statistics","recId":"538998830","language":["eng"]}],"physDesc":[{"extent":"47 S."}],"name":{"displayForm":["Joseph T. Meyer"]},"id":{"doi":["10.1214/23-EJS2187"],"eki":["1885863268"]},"origin":[{"dateIssuedKey":"2023","dateIssuedDisp":"7 December 2023"}]} | ||
| SRT | |a MEYERJOSEPOPTIMALCON7202 | ||