Resource-efficient neural networks for embedded systems
While machine learning is traditionally a resource intensive task, embedded systems, autonomous navigation, and the vision of the Internet of Things fuel the interest in resourceefficient approaches. These approaches aim for a carefully chosen trade-off between performance and resource consumption i...
Saved in:
| Main Authors: | , , , , , , , |
|---|---|
| Format: | Article (Journal) |
| Language: | English |
| Published: |
2024
|
| In: |
Journal of machine learning research
Year: 2024, Volume: 25, Pages: 1-51 |
| ISSN: | 1533-7928 |
| Online Access: | Verlag, lizenzpflichtig, Volltext: https://www.jmlr.org/papers/volume25/18-566/18-566.pdf
|
| Author Notes: | Wolfgang Roth (Graz University of Technology, Austria Laboratory of Signal Processing and Speech Communication), Günther Schindler (Heidelberg University, Germany Institute of Computer Engineering), Bernhard Klein (Heidelberg University, Germany Institute of Computer Engineering), Robert Peharz (Graz University of Technology, Austria Institute of Theoretical Computer Science), Sebastian Tschiatschek (University of Vienna, Austria Faculty of Computer Science), Holger Fröning (Heidelberg University, Germany Institute of Computer Engineering), Franz Pernkopf (Graz University of Technology, Austria Laboratory of Signal Processing and Speech Communication), Zoubin Ghahramani (University of Cambridge, UK) ; editor: Russ Greiner |
MARC
| LEADER | 00000caa a2200000 c 4500 | ||
|---|---|---|---|
| 001 | 1902317254 | ||
| 003 | DE-627 | ||
| 005 | 20241205171412.0 | ||
| 007 | cr uuu---uuuuu | ||
| 008 | 240911s2024 xx |||||o 00| ||eng c | ||
| 035 | |a (DE-627)1902317254 | ||
| 035 | |a (DE-599)KXP1902317254 | ||
| 035 | |a (OCoLC)1475311294 | ||
| 040 | |a DE-627 |b ger |c DE-627 |e rda | ||
| 041 | |a eng | ||
| 084 | |a 28 |2 sdnb | ||
| 100 | 1 | |a Roth, Wolfgang |e VerfasserIn |0 (DE-588)1263780008 |0 (DE-627)1811877818 |4 aut | |
| 245 | 1 | 0 | |a Resource-efficient neural networks for embedded systems |c Wolfgang Roth (Graz University of Technology, Austria Laboratory of Signal Processing and Speech Communication), Günther Schindler (Heidelberg University, Germany Institute of Computer Engineering), Bernhard Klein (Heidelberg University, Germany Institute of Computer Engineering), Robert Peharz (Graz University of Technology, Austria Institute of Theoretical Computer Science), Sebastian Tschiatschek (University of Vienna, Austria Faculty of Computer Science), Holger Fröning (Heidelberg University, Germany Institute of Computer Engineering), Franz Pernkopf (Graz University of Technology, Austria Laboratory of Signal Processing and Speech Communication), Zoubin Ghahramani (University of Cambridge, UK) ; editor: Russ Greiner |
| 264 | 1 | |c 2024 | |
| 300 | |a 51 | ||
| 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 11.09.2024 | ||
| 520 | |a While machine learning is traditionally a resource intensive task, embedded systems, autonomous navigation, and the vision of the Internet of Things fuel the interest in resourceefficient approaches. These approaches aim for a carefully chosen trade-off between performance and resource consumption in terms of computation and energy. The development of such approaches is among the major challenges in current machine learning research and key to ensure a smooth transition of machine learning technology from a scientific environment with virtually unlimited computing resources into everyday's applications. In this article, we provide an overview of the current state of the art of machine learning techniques facilitating these real -world requirements. In particular, we focus on resource -efficient inference based on deep neural networks (DNNs), the predominant machine learning models of the past decade. We give a comprehensive overview of the vast literature that can be mainly split into three non -mutually exclusive categories: (i) quantized neural networks, (ii) network pruning, and (iii) structural efficiency. These techniques can be applied dur ing training or as post -processing, and they are widely used to reduce the computational demands in terms of memory footprint, inference speed, and energy efficiency. We also briefly discuss different concepts of embedded hardware for DNNs and their compatibility with machine learning techniques as well as potential for energy and latency reduction. We substantiate our discussion with experiments on well-known benchmark data sets using compression techniques (quantization, pruning) for a set of resource -constrained embedded systems, such as CPUs, GPUs and FPGAs. The obtained results highlight the difficulty of finding good trade-offs between resource efficiency and prediction quality. | ||
| 650 | 4 | |a deep neural networks | |
| 650 | 4 | |a inference | |
| 650 | 4 | |a Resource-efficient machine learning | |
| 700 | 1 | |a Schindler, Günther |d 1990- |e VerfasserIn |0 (DE-588)1237058287 |0 (DE-627)1762956160 |4 aut | |
| 700 | 1 | |a Klein, Bernhard |e VerfasserIn |0 (DE-588)1341625443 |0 (DE-627)1902317734 |4 aut | |
| 700 | 1 | |a Peharz, Robert |e VerfasserIn |4 aut | |
| 700 | 1 | |a Tschiatschek, Sebastian |e VerfasserIn |4 aut | |
| 700 | 1 | |a Fröning, Holger |d 1976- |e VerfasserIn |0 (DE-588)133209466 |0 (DE-627)538678658 |0 (DE-576)299696189 |4 aut | |
| 700 | 1 | |a Pernkopf, Franz |e VerfasserIn |4 aut | |
| 700 | 1 | |a Ghahramani, Zoubin |e VerfasserIn |4 aut | |
| 773 | 0 | 8 | |i Enthalten in |t Journal of machine learning research |d Brookline, MA : Microtome Publishing, 2001 |g 25(2024), Seite 1-51 |h Online-Ressource |w (DE-627)327304472 |w (DE-600)2042762-1 |w (DE-576)112815766 |x 1533-7928 |7 nnas |a Resource-efficient neural networks for embedded systems |
| 773 | 1 | 8 | |g volume:25 |g year:2024 |g pages:1-51 |g extent:51 |a Resource-efficient neural networks for embedded systems |
| 856 | 4 | 0 | |u https://www.jmlr.org/papers/volume25/18-566/18-566.pdf |x Verlag |z lizenzpflichtig |3 Volltext |7 1 |
| 951 | |a AR | ||
| 992 | |a 20240911 | ||
| 993 | |a Article | ||
| 994 | |a 2024 | ||
| 998 | |g 133209466 |a Fröning, Holger |m 133209466:Fröning, Holger |d 700000 |d 720000 |e 700000PF133209466 |e 720000PF133209466 |k 0/700000/ |k 1/700000/720000/ |p 6 | ||
| 998 | |g 1341625443 |a Klein, Bernhard |m 1341625443:Klein, Bernhard |d 700000 |d 720000 |d 110000 |e 700000PK1341625443 |e 720000PK1341625443 |e 110000PK1341625443 |k 0/700000/ |k 1/700000/720000/ |k 0/110000/ |p 3 | ||
| 998 | |g 1237058287 |a Schindler, Günther |m 1237058287:Schindler, Günther |d 700000 |d 720000 |e 700000PS1237058287 |e 720000PS1237058287 |k 0/700000/ |k 1/700000/720000/ |p 2 | ||
| 999 | |a KXP-PPN1902317254 |e 4578673987 | ||
| BIB | |a Y | ||
| SER | |a journal | ||
| JSO | |a {"person":[{"display":"Roth, Wolfgang","given":"Wolfgang","role":"aut","family":"Roth"},{"family":"Schindler","role":"aut","display":"Schindler, Günther","given":"Günther"},{"given":"Bernhard","display":"Klein, Bernhard","family":"Klein","role":"aut"},{"given":"Robert","display":"Peharz, Robert","family":"Peharz","role":"aut"},{"display":"Tschiatschek, Sebastian","given":"Sebastian","family":"Tschiatschek","role":"aut"},{"display":"Fröning, Holger","given":"Holger","family":"Fröning","role":"aut"},{"display":"Pernkopf, Franz","given":"Franz","family":"Pernkopf","role":"aut"},{"family":"Ghahramani","role":"aut","display":"Ghahramani, Zoubin","given":"Zoubin"}],"recId":"1902317254","note":["Gesehen am 11.09.2024"],"name":{"displayForm":["Wolfgang Roth (Graz University of Technology, Austria Laboratory of Signal Processing and Speech Communication), Günther Schindler (Heidelberg University, Germany Institute of Computer Engineering), Bernhard Klein (Heidelberg University, Germany Institute of Computer Engineering), Robert Peharz (Graz University of Technology, Austria Institute of Theoretical Computer Science), Sebastian Tschiatschek (University of Vienna, Austria Faculty of Computer Science), Holger Fröning (Heidelberg University, Germany Institute of Computer Engineering), Franz Pernkopf (Graz University of Technology, Austria Laboratory of Signal Processing and Speech Communication), Zoubin Ghahramani (University of Cambridge, UK) ; editor: Russ Greiner"]},"language":["eng"],"relHost":[{"origin":[{"dateIssuedKey":"2001","publisherPlace":"Brookline, MA ; New York, NY ; Cambridge, Mass.","publisher":"Microtome Publishing ; Association for Computing Machinery ; MIT Press","dateIssuedDisp":"2001-"}],"physDesc":[{"extent":"Online-Ressource"}],"id":{"eki":["327304472"],"issn":["1533-7928"],"zdb":["2042762-1"]},"title":[{"subtitle":"JMLR","title":"Journal of machine learning research","title_sort":"Journal of machine learning research"}],"recId":"327304472","titleAlt":[{"title":"JMLR"}],"disp":"Resource-efficient neural networks for embedded systemsJournal of machine learning research","type":{"bibl":"periodical","media":"Online-Ressource"},"pubHistory":["1.2001 -"],"note":["Gesehen am 14.06.2023","Fortsetzung der Druck-Ausgabe"],"part":{"text":"25(2024), Seite 1-51","pages":"1-51","extent":"51","volume":"25","year":"2024"},"language":["eng"]}],"origin":[{"dateIssuedDisp":"2024","dateIssuedKey":"2024"}],"title":[{"title":"Resource-efficient neural networks for embedded systems","title_sort":"Resource-efficient neural networks for embedded systems"}],"type":{"bibl":"article-journal","media":"Online-Ressource"},"physDesc":[{"extent":"51 S."}],"id":{"eki":["1902317254"]}} | ||
| SRT | |a ROTHWOLFGARESOURCEEF2024 | ||