An Overview of Some Issues in the Theory of Deep Networks
© 2020 Institute of Electrical Engineers of Japan. Published by Wiley Periodicals LLC. During the last few years, significant progress has been made in the theoretical understanding of deep networks. We review our contributions in the areas of approximation theory and optimization. We also introduce...
| Main Authors: | , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
Wiley
2021
|
| Online Access: | https://hdl.handle.net/1721.1/138407 |
| _version_ | 1811074238214307840 |
|---|---|
| author | Poggio, Tomaso Banburski, Andrzej |
| author_facet | Poggio, Tomaso Banburski, Andrzej |
| author_sort | Poggio, Tomaso |
| collection | MIT |
| description | © 2020 Institute of Electrical Engineers of Japan. Published by Wiley Periodicals LLC. During the last few years, significant progress has been made in the theoretical understanding of deep networks. We review our contributions in the areas of approximation theory and optimization. We also introduce a new approach based on cross-validation leave-one-out stability to estimate bounds on the expected error of overparametrized classifiers, such as deep networks. © 2020 Institute of Electrical Engineers of Japan. Published by Wiley Periodicals LLC. |
| first_indexed | 2024-09-23T09:46:09Z |
| format | Article |
| id | mit-1721.1/138407 |
| institution | Massachusetts Institute of Technology |
| language | English |
| last_indexed | 2024-09-23T09:46:09Z |
| publishDate | 2021 |
| publisher | Wiley |
| record_format | dspace |
| spelling | mit-1721.1/1384072021-12-10T03:31:25Z An Overview of Some Issues in the Theory of Deep Networks Poggio, Tomaso Banburski, Andrzej © 2020 Institute of Electrical Engineers of Japan. Published by Wiley Periodicals LLC. During the last few years, significant progress has been made in the theoretical understanding of deep networks. We review our contributions in the areas of approximation theory and optimization. We also introduce a new approach based on cross-validation leave-one-out stability to estimate bounds on the expected error of overparametrized classifiers, such as deep networks. © 2020 Institute of Electrical Engineers of Japan. Published by Wiley Periodicals LLC. 2021-12-09T19:16:21Z 2021-12-09T19:16:21Z 2020 2021-12-09T19:11:49Z Article http://purl.org/eprint/type/JournalArticle https://hdl.handle.net/1721.1/138407 Poggio, Tomaso and Banburski, Andrzej. 2020. "An Overview of Some Issues in the Theory of Deep Networks." IEEJ Transactions on Electrical and Electronic Engineering, 15 (11). en 10.1002/TEE.23243 IEEJ Transactions on Electrical and Electronic Engineering Creative Commons Attribution-Noncommercial-Share Alike http://creativecommons.org/licenses/by-nc-sa/4.0/ application/pdf Wiley Prof. Poggio |
| spellingShingle | Poggio, Tomaso Banburski, Andrzej An Overview of Some Issues in the Theory of Deep Networks |
| title | An Overview of Some Issues in the Theory of Deep Networks |
| title_full | An Overview of Some Issues in the Theory of Deep Networks |
| title_fullStr | An Overview of Some Issues in the Theory of Deep Networks |
| title_full_unstemmed | An Overview of Some Issues in the Theory of Deep Networks |
| title_short | An Overview of Some Issues in the Theory of Deep Networks |
| title_sort | overview of some issues in the theory of deep networks |
| url | https://hdl.handle.net/1721.1/138407 |
| work_keys_str_mv | AT poggiotomaso anoverviewofsomeissuesinthetheoryofdeepnetworks AT banburskiandrzej anoverviewofsomeissuesinthetheoryofdeepnetworks AT poggiotomaso overviewofsomeissuesinthetheoryofdeepnetworks AT banburskiandrzej overviewofsomeissuesinthetheoryofdeepnetworks |