ACE-CNN: approximate carry disregard multipliers for energy-efficient CNN-based image classification
This paper presents the design and development of Signed Carry Disregard Multiplier (SCDM8), a family of signed approximate multipliers tailored for integration into Convolutional Neural Networks (CNNs). Extensive experiments were conducted on popular pre-trained CNN models, including VGG16, VGG19,...
Gespeichert in:
| Hauptverfasser: | , , , , |
|---|---|
| Dokumenttyp: | Article (Journal) |
| Sprache: | Englisch |
| Veröffentlicht: |
May 2024
|
| In: |
IEEE transactions on circuits and systems. 1, Regular papers
Year: 2024, Jahrgang: 71, Heft: 5, Pages: 2280-2293 |
| ISSN: | 1558-0806 |
| DOI: | 10.1109/TCSI.2024.3369230 |
| Online-Zugang: | Verlag, kostenfrei, Volltext: https://doi.org/10.1109/TCSI.2024.3369230 Verlag, kostenfrei, Volltext: https://www.webofscience.com/api/gateway?GWVersion=2&SrcAuth=DynamicDOIArticle&SrcApp=WOS&KeyAID=10.1109%2FTCSI.2024.3369230&DestApp=DOI&SrcAppSID=EUW1ED0F43bzmle4M5CqqQBSdgNxH&SrcJTitle=IEEE+TRANSACTIONS+ON+CIRCUITS+AND+SYSTEMS+I-REGULAR+PAPERS&DestDOIRegistrantName=Institute+of+Electrical+and+Electronics+Engineers 
|
| Verfasserangaben: | Salar Shakibhamedan, Nima Amirafshar, Graduate Student Member, IEEE, Ahmad Sedigh Baroughi , Hadi Shahriar Shahhoseini, and Nima Taherinejad, Member, IEEE |
MARC
| LEADER | 00000caa a2200000 c 4500 | ||
|---|---|---|---|
| 001 | 1895988691 | ||
| 003 | DE-627 | ||
| 005 | 20241205153458.0 | ||
| 007 | cr uuu---uuuuu | ||
| 008 | 240722s2024 xx |||||o 00| ||eng c | ||
| 024 | 7 | |a 10.1109/TCSI.2024.3369230 |2 doi | |
| 035 | |a (DE-627)1895988691 | ||
| 035 | |a (DE-599)KXP1895988691 | ||
| 035 | |a (OCoLC)1475303588 | ||
| 040 | |a DE-627 |b ger |c DE-627 |e rda | ||
| 041 | |a eng | ||
| 084 | |a 28 |2 sdnb | ||
| 100 | 1 | |a Shakibhamedan, Salar |e VerfasserIn |0 (DE-588)1336458593 |0 (DE-627)1895989175 |4 aut | |
| 245 | 1 | 0 | |a ACE-CNN |b approximate carry disregard multipliers for energy-efficient CNN-based image classification |c Salar Shakibhamedan, Nima Amirafshar, Graduate Student Member, IEEE, Ahmad Sedigh Baroughi , Hadi Shahriar Shahhoseini, and Nima Taherinejad, Member, IEEE |
| 264 | 1 | |c May 2024 | |
| 300 | |b Illustrationen | ||
| 300 | |a 14 | ||
| 336 | |a Text |b txt |2 rdacontent | ||
| 337 | |a Computermedien |b c |2 rdamedia | ||
| 338 | |a Online-Ressource |b cr |2 rdacarrier | ||
| 500 | |a Veröffentlicht: 01. März 2024 | ||
| 500 | |a Gesehen am 22.07.2024 | ||
| 520 | |a This paper presents the design and development of Signed Carry Disregard Multiplier (SCDM8), a family of signed approximate multipliers tailored for integration into Convolutional Neural Networks (CNNs). Extensive experiments were conducted on popular pre-trained CNN models, including VGG16, VGG19, ResNet101, ResNet152, MobileNetV2, InceptionV3, and ConvNeXt-T to evaluate the trade-off between accuracy and approximation. The results demonstrate that ACE-CNN outperforms other configurations, offering a favorable balance between accuracy and computational efficiency. In our experiments, when applied to VGG16, SCDM8 achieves an average reduction in power consumption of 35% with a marginal decrease in accuracy of only 1.5%. Similarly, when incorporated into ResNet152, SCDM8 yields an energy saving of 42% while sacrificing only 1.8% in accuracy. ACE-CNN provides the first approximate version of ConvNeXt which yields up to 72% energy improvement at the price of less than only 1.3% Top-1 accuracy. These results highlight the suitability of SCDM8 as an approximation method across various CNN models. Our analysis shows that the ACE-CNN outperforms state-of-the-art approaches in accuracy, energy efficiency, and computation precision for image classification tasks in CNNs. Our study investigated the resiliency of CNN models to approximate multipliers, revealing that ResNet101 demonstrated the highest resiliency with an average difference in the accuracy of 0.97%, whereas LeNet5 Inspired-CNN exhibited the lowest resiliency with an average difference of 2.92%. These findings aid in selecting energy-efficient approximate multipliers for CNN-based systems, and contribute to the development of energy-efficient deep learning systems by offering an effective approximation technique for multipliers in CNNs. The proposed SCDM8 family of approximate multipliers opens new avenues for efficient deep learning applications, enabling significant energy savings with virtually no loss in accuracy. | ||
| 650 | 4 | |a 4-2 COMPRESSORS | |
| 650 | 4 | |a approximate multiplier | |
| 650 | 4 | |a Computer architecture | |
| 650 | 4 | |a convolutional neural network | |
| 650 | 4 | |a Convolutional neural networks | |
| 650 | 4 | |a Delays | |
| 650 | 4 | |a DESIGN | |
| 650 | 4 | |a Energy efficiency | |
| 650 | 4 | |a Hardware | |
| 650 | 4 | |a image classification | |
| 650 | 4 | |a Image classification | |
| 650 | 4 | |a Task analysis | |
| 700 | 1 | |a Amirafshar, Nima |e VerfasserIn |4 aut | |
| 700 | 1 | |a Baroughi, Ahmad Sedigh |e VerfasserIn |4 aut | |
| 700 | 1 | |a Shahhoseini, Hadi Shahriar |e VerfasserIn |4 aut | |
| 700 | 1 | |a Taherinejad, Nima |e VerfasserIn |0 (DE-588)127517549X |0 (DE-627)1826739688 |4 aut | |
| 773 | 0 | 8 | |i Enthalten in |a Institute of Electrical and Electronics Engineers |t IEEE transactions on circuits and systems. 1, Regular papers |d New York, NY : Institute of Electrical and Electronics Engineers, 2004 |g 71(2024), 5 vom: Mai, Seite 2280-2293 |h Online-Ressource |w (DE-627)324532180 |w (DE-600)2028230-8 |w (DE-576)094080798 |x 1558-0806 |7 nnas |
| 773 | 1 | 8 | |g volume:71 |g year:2024 |g number:5 |g month:05 |g pages:2280-2293 |g extent:14 |a ACE-CNN approximate carry disregard multipliers for energy-efficient CNN-based image classification |
| 856 | 4 | 0 | |u https://doi.org/10.1109/TCSI.2024.3369230 |x Verlag |x Resolving-System |z kostenfrei |3 Volltext |
| 856 | 4 | 0 | |u https://www.webofscience.com/api/gateway?GWVersion=2&SrcAuth=DynamicDOIArticle&SrcApp=WOS&KeyAID=10.1109%2FTCSI.2024.3369230&DestApp=DOI&SrcAppSID=EUW1ED0F43bzmle4M5CqqQBSdgNxH&SrcJTitle=IEEE+TRANSACTIONS+ON+CIRCUITS+AND+SYSTEMS+I-REGULAR+PAPERS&DestDOIRegistrantName=Institute+of+Electrical+and+Electronics+Engineers |x Verlag |z kostenfrei |3 Volltext |
| 951 | |a AR | ||
| 992 | |a 20240722 | ||
| 993 | |a Article | ||
| 994 | |a 2024 | ||
| 998 | |g 127517549X |a Taherinejad, Nima |m 127517549X:Taherinejad, Nima |d 700000 |d 720000 |e 700000PT127517549X |e 720000PT127517549X |k 0/700000/ |k 1/700000/720000/ |p 5 |y j | ||
| 999 | |a KXP-PPN1895988691 |e 4555316509 | ||
| BIB | |a Y | ||
| SER | |a journal | ||
| JSO | |a {"language":["eng"],"title":[{"title_sort":"ACE-CNN","title":"ACE-CNN","subtitle":"approximate carry disregard multipliers for energy-efficient CNN-based image classification"}],"person":[{"family":"Shakibhamedan","given":"Salar","role":"aut","display":"Shakibhamedan, Salar"},{"display":"Amirafshar, Nima","role":"aut","given":"Nima","family":"Amirafshar"},{"role":"aut","given":"Ahmad Sedigh","display":"Baroughi, Ahmad Sedigh","family":"Baroughi"},{"family":"Shahhoseini","role":"aut","given":"Hadi Shahriar","display":"Shahhoseini, Hadi Shahriar"},{"display":"Taherinejad, Nima","given":"Nima","role":"aut","family":"Taherinejad"}],"name":{"displayForm":["Salar Shakibhamedan, Nima Amirafshar, Graduate Student Member, IEEE, Ahmad Sedigh Baroughi , Hadi Shahriar Shahhoseini, and Nima Taherinejad, Member, IEEE"]},"relHost":[{"language":["eng"],"part":{"text":"71(2024), 5 vom: Mai, Seite 2280-2293","year":"2024","volume":"71","pages":"2280-2293","extent":"14","issue":"5"},"title":[{"title_sort":"IEEE transactions on circuits and systems","subtitle":"a publication of the IEEE Circuits and Systems Society","title":"IEEE transactions on circuits and systems","partname":"Regular papers"}],"pubHistory":["Volume 51, number 1 (January 2004)-"],"origin":[{"dateIssuedKey":"2004","publisher":"Institute of Electrical and Electronics Engineers","dateIssuedDisp":"2004-","publisherPlace":"New York, NY"}],"recId":"324532180","titleAlt":[{"title":"IEEE transactions on circuits and systems / 1"},{"title":"Transactions on circuits and systems"}],"type":{"media":"Online-Ressource","bibl":"periodical"},"note":["Gesehen am 18.08.2022"],"corporate":[{"role":"aut","display":"Institute of Electrical and Electronics Engineers"},{"role":"isb","display":"IEEE Circuits and Systems Society"}],"disp":"Institute of Electrical and Electronics EngineersIEEE transactions on circuits and systems. 1, Regular papers","physDesc":[{"extent":"Online-Ressource"}],"id":{"eki":["324532180"],"issn":["1558-0806"],"zdb":["2028230-8"],"doi":["10.1109/TCSI.8919"]}}],"note":["Veröffentlicht: 01. März 2024","Gesehen am 22.07.2024"],"physDesc":[{"extent":"14 S.","noteIll":"Illustrationen"}],"id":{"eki":["1895988691"],"doi":["10.1109/TCSI.2024.3369230"]},"origin":[{"dateIssuedKey":"2024","dateIssuedDisp":"May 2024"}],"recId":"1895988691","type":{"bibl":"article-journal","media":"Online-Ressource"}} | ||
| SRT | |a SHAKIBHAMEACECNN2024 | ||