Photonic Binary Convolutional Neural Network Based on Microring Resonator Array
We propose the photonic architecture based on microring resonator (MRR) arrays for binary convolutional neural networks (BCNNs) accelerated computing. The MRR crossbar unit is used for computing weight {-1, 1} and the single MRR is for input {0, 1}. The computing parallelism is improved through wave...
Gespeichert in:
| Veröffentlicht in: | IEEE photonics technology letters 2023-06, Vol.35 (12), p.1-1 |
|---|---|
| Hauptverfasser: | , , , , , , , |
| Format: | Artikel |
| Sprache: | eng |
| Schlagworte: | |
| Online-Zugang: | Volltext bestellen |
| Tags: |
Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
|
| container_end_page | 1 |
|---|---|
| container_issue | 12 |
| container_start_page | 1 |
| container_title | IEEE photonics technology letters |
| container_volume | 35 |
| creator | Wang, Ruiting Wang, Pengfei Lyu, Chen Luo, Guangzhen Ma, Jianbin Zhou, Xuliang Zhang, Yejin Pan, Jiaoqing |
| description | We propose the photonic architecture based on microring resonator (MRR) arrays for binary convolutional neural networks (BCNNs) accelerated computing. The MRR crossbar unit is used for computing weight {-1, 1} and the single MRR is for input {0, 1}. The computing parallelism is improved through wavelength division multiplexing. The photonic BCNN achieves 97.29% classification accuracy on the MNIST test set which is only 1.94% lower than the accuracy of the 32-bit neural network, and saves 32× at memory usage. We analyze effects of input and weight encoding errors on the photonic BCNN. When the input or weight error rate is less than 0.01%, the test accuracy remains unchanged. We evaluate the performance of the photonic BCNN architecture considering optical loss, inter-channel crosstalk, operation frequency and device power consumption. The energy efficiency of the designed photonic BCNN architecture is 1.72 pJ/MAC, which is 4.80× and 61.32× better than the 8-bit and 16-bit architecture respectively. The photonic BCNN is promising to be used for edge computing. |
| doi_str_mv | 10.1109/LPT.2023.3272148 |
| format | Article |
| fullrecord | <record><control><sourceid>proquest_RIE</sourceid><recordid>TN_cdi_proquest_journals_2811730536</recordid><sourceformat>XML</sourceformat><sourcesystem>PC</sourcesystem><ieee_id>10113659</ieee_id><sourcerecordid>2811730536</sourcerecordid><originalsourceid>FETCH-LOGICAL-c292t-5ca8ad188b5ea339ca4b94bd4956afef63a46cd68badebdb6046944ef17b7af63</originalsourceid><addsrcrecordid>eNpNkEtPwzAQhC0EEqVw58DBEucUr1-Jj23FSyq0QuVs2YkDKSUudgLqv8elPXCakXZmtfshdAlkBEDUzWyxHFFC2YjRnAIvjtAAFIeMQM6PkyfJAzBxis5iXBECXDA-QPPFu-9825R40rQmbPHUt99-3XeNb80aP7s-_En348MHnpjoKuxb_NSUwYemfcMvLqZk5wMeh2C25-ikNuvoLg46RK93t8vpQzab3z9Ox7OspIp2mShNYSooCiucYUyVhlvFbcWVkKZ2tWSGy7KShTWVs5WVhEvFuasht7lJ4yG63u_dBP_Vu9jple9DOjlqWgDkjAi2S5F9Kl0bY3C13oTmM72pgegdNp2w6R02fcCWKlf7SuOc-xdP7KRQ7Bcj1GoW</addsrcrecordid><sourcetype>Aggregation Database</sourcetype><iscdi>true</iscdi><recordtype>article</recordtype><pqid>2811730536</pqid></control><display><type>article</type><title>Photonic Binary Convolutional Neural Network Based on Microring Resonator Array</title><source>IEEE Electronic Library (IEL)</source><creator>Wang, Ruiting ; Wang, Pengfei ; Lyu, Chen ; Luo, Guangzhen ; Ma, Jianbin ; Zhou, Xuliang ; Zhang, Yejin ; Pan, Jiaoqing</creator><creatorcontrib>Wang, Ruiting ; Wang, Pengfei ; Lyu, Chen ; Luo, Guangzhen ; Ma, Jianbin ; Zhou, Xuliang ; Zhang, Yejin ; Pan, Jiaoqing</creatorcontrib><description>We propose the photonic architecture based on microring resonator (MRR) arrays for binary convolutional neural networks (BCNNs) accelerated computing. The MRR crossbar unit is used for computing weight {-1, 1} and the single MRR is for input {0, 1}. The computing parallelism is improved through wavelength division multiplexing. The photonic BCNN achieves 97.29% classification accuracy on the MNIST test set which is only 1.94% lower than the accuracy of the 32-bit neural network, and saves 32× at memory usage. We analyze effects of input and weight encoding errors on the photonic BCNN. When the input or weight error rate is less than 0.01%, the test accuracy remains unchanged. We evaluate the performance of the photonic BCNN architecture considering optical loss, inter-channel crosstalk, operation frequency and device power consumption. The energy efficiency of the designed photonic BCNN architecture is 1.72 pJ/MAC, which is 4.80× and 61.32× better than the 8-bit and 16-bit architecture respectively. The photonic BCNN is promising to be used for edge computing.</description><identifier>ISSN: 1041-1135</identifier><identifier>EISSN: 1941-0174</identifier><identifier>DOI: 10.1109/LPT.2023.3272148</identifier><identifier>CODEN: IPTLEL</identifier><language>eng</language><publisher>New York: IEEE</publisher><subject>Accuracy ; Arrays ; Artificial neural networks ; binary neural network ; Computer architecture ; Computer networks ; Convolution ; Crosstalk ; Edge computing ; Kernel ; microring resonator array ; Neural networks ; Optical losses ; Optical resonators ; Optical waveguides ; Photonic neural network ; Photonics ; Power consumption ; Resonators ; Wavelength division multiplexing</subject><ispartof>IEEE photonics technology letters, 2023-06, Vol.35 (12), p.1-1</ispartof><rights>Copyright The Institute of Electrical and Electronics Engineers, Inc. (IEEE) 2023</rights><woscitedreferencessubscribed>false</woscitedreferencessubscribed><citedby>FETCH-LOGICAL-c292t-5ca8ad188b5ea339ca4b94bd4956afef63a46cd68badebdb6046944ef17b7af63</citedby><cites>FETCH-LOGICAL-c292t-5ca8ad188b5ea339ca4b94bd4956afef63a46cd68badebdb6046944ef17b7af63</cites><orcidid>0000-0001-9874-3803 ; 0000-0003-4701-9561 ; 0000-0001-7426-199X ; 0000-0001-9932-9372 ; 0000-0002-2306-0730</orcidid></display><links><openurl>$$Topenurl_article</openurl><openurlfulltext>$$Topenurlfull_article</openurlfulltext><thumbnail>$$Tsyndetics_thumb_exl</thumbnail><linktohtml>$$Uhttps://ieeexplore.ieee.org/document/10113659$$EHTML$$P50$$Gieee$$H</linktohtml><link.rule.ids>314,776,780,792,27901,27902,54733</link.rule.ids><linktorsrc>$$Uhttps://ieeexplore.ieee.org/document/10113659$$EView_record_in_IEEE$$FView_record_in_$$GIEEE</linktorsrc></links><search><creatorcontrib>Wang, Ruiting</creatorcontrib><creatorcontrib>Wang, Pengfei</creatorcontrib><creatorcontrib>Lyu, Chen</creatorcontrib><creatorcontrib>Luo, Guangzhen</creatorcontrib><creatorcontrib>Ma, Jianbin</creatorcontrib><creatorcontrib>Zhou, Xuliang</creatorcontrib><creatorcontrib>Zhang, Yejin</creatorcontrib><creatorcontrib>Pan, Jiaoqing</creatorcontrib><title>Photonic Binary Convolutional Neural Network Based on Microring Resonator Array</title><title>IEEE photonics technology letters</title><addtitle>LPT</addtitle><description>We propose the photonic architecture based on microring resonator (MRR) arrays for binary convolutional neural networks (BCNNs) accelerated computing. The MRR crossbar unit is used for computing weight {-1, 1} and the single MRR is for input {0, 1}. The computing parallelism is improved through wavelength division multiplexing. The photonic BCNN achieves 97.29% classification accuracy on the MNIST test set which is only 1.94% lower than the accuracy of the 32-bit neural network, and saves 32× at memory usage. We analyze effects of input and weight encoding errors on the photonic BCNN. When the input or weight error rate is less than 0.01%, the test accuracy remains unchanged. We evaluate the performance of the photonic BCNN architecture considering optical loss, inter-channel crosstalk, operation frequency and device power consumption. The energy efficiency of the designed photonic BCNN architecture is 1.72 pJ/MAC, which is 4.80× and 61.32× better than the 8-bit and 16-bit architecture respectively. The photonic BCNN is promising to be used for edge computing.</description><subject>Accuracy</subject><subject>Arrays</subject><subject>Artificial neural networks</subject><subject>binary neural network</subject><subject>Computer architecture</subject><subject>Computer networks</subject><subject>Convolution</subject><subject>Crosstalk</subject><subject>Edge computing</subject><subject>Kernel</subject><subject>microring resonator array</subject><subject>Neural networks</subject><subject>Optical losses</subject><subject>Optical resonators</subject><subject>Optical waveguides</subject><subject>Photonic neural network</subject><subject>Photonics</subject><subject>Power consumption</subject><subject>Resonators</subject><subject>Wavelength division multiplexing</subject><issn>1041-1135</issn><issn>1941-0174</issn><fulltext>true</fulltext><rsrctype>article</rsrctype><creationdate>2023</creationdate><recordtype>article</recordtype><sourceid>RIE</sourceid><recordid>eNpNkEtPwzAQhC0EEqVw58DBEucUr1-Jj23FSyq0QuVs2YkDKSUudgLqv8elPXCakXZmtfshdAlkBEDUzWyxHFFC2YjRnAIvjtAAFIeMQM6PkyfJAzBxis5iXBECXDA-QPPFu-9825R40rQmbPHUt99-3XeNb80aP7s-_En348MHnpjoKuxb_NSUwYemfcMvLqZk5wMeh2C25-ikNuvoLg46RK93t8vpQzab3z9Ox7OspIp2mShNYSooCiucYUyVhlvFbcWVkKZ2tWSGy7KShTWVs5WVhEvFuasht7lJ4yG63u_dBP_Vu9jple9DOjlqWgDkjAi2S5F9Kl0bY3C13oTmM72pgegdNp2w6R02fcCWKlf7SuOc-xdP7KRQ7Bcj1GoW</recordid><startdate>20230615</startdate><enddate>20230615</enddate><creator>Wang, Ruiting</creator><creator>Wang, Pengfei</creator><creator>Lyu, Chen</creator><creator>Luo, Guangzhen</creator><creator>Ma, Jianbin</creator><creator>Zhou, Xuliang</creator><creator>Zhang, Yejin</creator><creator>Pan, Jiaoqing</creator><general>IEEE</general><general>The Institute of Electrical and Electronics Engineers, Inc. (IEEE)</general><scope>97E</scope><scope>RIA</scope><scope>RIE</scope><scope>AAYXX</scope><scope>CITATION</scope><scope>7SP</scope><scope>7U5</scope><scope>8FD</scope><scope>L7M</scope><orcidid>https://orcid.org/0000-0001-9874-3803</orcidid><orcidid>https://orcid.org/0000-0003-4701-9561</orcidid><orcidid>https://orcid.org/0000-0001-7426-199X</orcidid><orcidid>https://orcid.org/0000-0001-9932-9372</orcidid><orcidid>https://orcid.org/0000-0002-2306-0730</orcidid></search><sort><creationdate>20230615</creationdate><title>Photonic Binary Convolutional Neural Network Based on Microring Resonator Array</title><author>Wang, Ruiting ; Wang, Pengfei ; Lyu, Chen ; Luo, Guangzhen ; Ma, Jianbin ; Zhou, Xuliang ; Zhang, Yejin ; Pan, Jiaoqing</author></sort><facets><frbrtype>5</frbrtype><frbrgroupid>cdi_FETCH-LOGICAL-c292t-5ca8ad188b5ea339ca4b94bd4956afef63a46cd68badebdb6046944ef17b7af63</frbrgroupid><rsrctype>articles</rsrctype><prefilter>articles</prefilter><language>eng</language><creationdate>2023</creationdate><topic>Accuracy</topic><topic>Arrays</topic><topic>Artificial neural networks</topic><topic>binary neural network</topic><topic>Computer architecture</topic><topic>Computer networks</topic><topic>Convolution</topic><topic>Crosstalk</topic><topic>Edge computing</topic><topic>Kernel</topic><topic>microring resonator array</topic><topic>Neural networks</topic><topic>Optical losses</topic><topic>Optical resonators</topic><topic>Optical waveguides</topic><topic>Photonic neural network</topic><topic>Photonics</topic><topic>Power consumption</topic><topic>Resonators</topic><topic>Wavelength division multiplexing</topic><toplevel>online_resources</toplevel><creatorcontrib>Wang, Ruiting</creatorcontrib><creatorcontrib>Wang, Pengfei</creatorcontrib><creatorcontrib>Lyu, Chen</creatorcontrib><creatorcontrib>Luo, Guangzhen</creatorcontrib><creatorcontrib>Ma, Jianbin</creatorcontrib><creatorcontrib>Zhou, Xuliang</creatorcontrib><creatorcontrib>Zhang, Yejin</creatorcontrib><creatorcontrib>Pan, Jiaoqing</creatorcontrib><collection>IEEE All-Society Periodicals Package (ASPP) 2005-present</collection><collection>IEEE All-Society Periodicals Package (ASPP) 1998-Present</collection><collection>IEEE Electronic Library (IEL)</collection><collection>CrossRef</collection><collection>Electronics & Communications Abstracts</collection><collection>Solid State and Superconductivity Abstracts</collection><collection>Technology Research Database</collection><collection>Advanced Technologies Database with Aerospace</collection><jtitle>IEEE photonics technology letters</jtitle></facets><delivery><delcategory>Remote Search Resource</delcategory><fulltext>fulltext_linktorsrc</fulltext></delivery><addata><au>Wang, Ruiting</au><au>Wang, Pengfei</au><au>Lyu, Chen</au><au>Luo, Guangzhen</au><au>Ma, Jianbin</au><au>Zhou, Xuliang</au><au>Zhang, Yejin</au><au>Pan, Jiaoqing</au><format>journal</format><genre>article</genre><ristype>JOUR</ristype><atitle>Photonic Binary Convolutional Neural Network Based on Microring Resonator Array</atitle><jtitle>IEEE photonics technology letters</jtitle><stitle>LPT</stitle><date>2023-06-15</date><risdate>2023</risdate><volume>35</volume><issue>12</issue><spage>1</spage><epage>1</epage><pages>1-1</pages><issn>1041-1135</issn><eissn>1941-0174</eissn><coden>IPTLEL</coden><abstract>We propose the photonic architecture based on microring resonator (MRR) arrays for binary convolutional neural networks (BCNNs) accelerated computing. The MRR crossbar unit is used for computing weight {-1, 1} and the single MRR is for input {0, 1}. The computing parallelism is improved through wavelength division multiplexing. The photonic BCNN achieves 97.29% classification accuracy on the MNIST test set which is only 1.94% lower than the accuracy of the 32-bit neural network, and saves 32× at memory usage. We analyze effects of input and weight encoding errors on the photonic BCNN. When the input or weight error rate is less than 0.01%, the test accuracy remains unchanged. We evaluate the performance of the photonic BCNN architecture considering optical loss, inter-channel crosstalk, operation frequency and device power consumption. The energy efficiency of the designed photonic BCNN architecture is 1.72 pJ/MAC, which is 4.80× and 61.32× better than the 8-bit and 16-bit architecture respectively. The photonic BCNN is promising to be used for edge computing.</abstract><cop>New York</cop><pub>IEEE</pub><doi>10.1109/LPT.2023.3272148</doi><tpages>1</tpages><orcidid>https://orcid.org/0000-0001-9874-3803</orcidid><orcidid>https://orcid.org/0000-0003-4701-9561</orcidid><orcidid>https://orcid.org/0000-0001-7426-199X</orcidid><orcidid>https://orcid.org/0000-0001-9932-9372</orcidid><orcidid>https://orcid.org/0000-0002-2306-0730</orcidid></addata></record> |
| fulltext | fulltext_linktorsrc |
| identifier | ISSN: 1041-1135 |
| ispartof | IEEE photonics technology letters, 2023-06, Vol.35 (12), p.1-1 |
| issn | 1041-1135 1941-0174 |
| language | eng |
| recordid | cdi_proquest_journals_2811730536 |
| source | IEEE Electronic Library (IEL) |
| subjects | Accuracy Arrays Artificial neural networks binary neural network Computer architecture Computer networks Convolution Crosstalk Edge computing Kernel microring resonator array Neural networks Optical losses Optical resonators Optical waveguides Photonic neural network Photonics Power consumption Resonators Wavelength division multiplexing |
| title | Photonic Binary Convolutional Neural Network Based on Microring Resonator Array |
| url | https://sfx.bib-bvb.de/sfx_tum?ctx_ver=Z39.88-2004&ctx_enc=info:ofi/enc:UTF-8&ctx_tim=2025-02-10T02%3A38%3A50IST&url_ver=Z39.88-2004&url_ctx_fmt=infofi/fmt:kev:mtx:ctx&rfr_id=info:sid/primo.exlibrisgroup.com:primo3-Article-proquest_RIE&rft_val_fmt=info:ofi/fmt:kev:mtx:journal&rft.genre=article&rft.atitle=Photonic%20Binary%20Convolutional%20Neural%20Network%20Based%20on%20Microring%20Resonator%20Array&rft.jtitle=IEEE%20photonics%20technology%20letters&rft.au=Wang,%20Ruiting&rft.date=2023-06-15&rft.volume=35&rft.issue=12&rft.spage=1&rft.epage=1&rft.pages=1-1&rft.issn=1041-1135&rft.eissn=1941-0174&rft.coden=IPTLEL&rft_id=info:doi/10.1109/LPT.2023.3272148&rft_dat=%3Cproquest_RIE%3E2811730536%3C/proquest_RIE%3E%3Curl%3E%3C/url%3E&disable_directlink=true&sfx.directlink=off&sfx.report_link=0&rft_id=info:oai/&rft_pqid=2811730536&rft_id=info:pmid/&rft_ieee_id=10113659&rfr_iscdi=true |