Purely self-rectifying memristor-based passive crossbar array for artificial neural network accelerators
Abstract Memristor-integrated passive crossbar arrays (CAs) could potentially accelerate neural network (NN) computations, but studies on these devices are limited to software-based simulations owing to their poor reliability. Herein, we propose a self-rectifying memristor-based 1 kb CA as a hardwar...
| Main Authors: | , , , , , , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
Nature Portfolio
2024-01-01
|
| Series: | Nature Communications |
| Online Access: | https://doi.org/10.1038/s41467-023-44620-1 |
| _version_ | 1797363145971335168 |
|---|---|
| author | Kanghyeok Jeon Jin Joo Ryu Seongil Im Hyun Kyu Seo Taeyong Eom Hyunsu Ju Min Kyu Yang Doo Seok Jeong Gun Hwan Kim |
| author_facet | Kanghyeok Jeon Jin Joo Ryu Seongil Im Hyun Kyu Seo Taeyong Eom Hyunsu Ju Min Kyu Yang Doo Seok Jeong Gun Hwan Kim |
| author_sort | Kanghyeok Jeon |
| collection | DOAJ |
| description | Abstract Memristor-integrated passive crossbar arrays (CAs) could potentially accelerate neural network (NN) computations, but studies on these devices are limited to software-based simulations owing to their poor reliability. Herein, we propose a self-rectifying memristor-based 1 kb CA as a hardware accelerator for NN computations. We conducted fully hardware-based single-layer NN classification tasks involving the Modified National Institute of Standards and Technology database using the developed passive CA, and achieved 100% classification accuracy for 1500 test sets. We also investigated the influences of the defect-tolerance capability of the CA, impact of the conductance range of the integrated memristors, and presence or absence of selection functionality in the integrated memristors on the image classification tasks. We offer valuable insights into the behavior and performance of CA devices under various conditions and provide evidence of the practicality of memristor-integrated passive CAs as hardware accelerators for NN applications. |
| first_indexed | 2024-03-08T16:16:18Z |
| format | Article |
| id | doaj.art-64286c995cc645499a041ed1815c8295 |
| institution | Directory Open Access Journal |
| issn | 2041-1723 |
| language | English |
| last_indexed | 2024-03-08T16:16:18Z |
| publishDate | 2024-01-01 |
| publisher | Nature Portfolio |
| record_format | Article |
| series | Nature Communications |
| spelling | doaj.art-64286c995cc645499a041ed1815c82952024-01-07T12:33:13ZengNature PortfolioNature Communications2041-17232024-01-0115111310.1038/s41467-023-44620-1Purely self-rectifying memristor-based passive crossbar array for artificial neural network acceleratorsKanghyeok Jeon0Jin Joo Ryu1Seongil Im2Hyun Kyu Seo3Taeyong Eom4Hyunsu Ju5Min Kyu Yang6Doo Seok Jeong7Gun Hwan Kim8Division of Materials Science and Engineering, Hanyang UniversityDivision of Advanced Materials, Korea Research Institute of Chemical Technology (KRICT)Center for Opto-Electronic Materials and Devices, Korea Institute of Science and Technology (KIST)Intelligent Electronic Device Lab, Sahmyook UniversityDivision of Advanced Materials, Korea Research Institute of Chemical Technology (KRICT)Center for Opto-Electronic Materials and Devices, Korea Institute of Science and Technology (KIST)Intelligent Electronic Device Lab, Sahmyook UniversityDivision of Materials Science and Engineering, Hanyang UniversityDepartment of Materials Science and Engineering, Yonsei UniversityAbstract Memristor-integrated passive crossbar arrays (CAs) could potentially accelerate neural network (NN) computations, but studies on these devices are limited to software-based simulations owing to their poor reliability. Herein, we propose a self-rectifying memristor-based 1 kb CA as a hardware accelerator for NN computations. We conducted fully hardware-based single-layer NN classification tasks involving the Modified National Institute of Standards and Technology database using the developed passive CA, and achieved 100% classification accuracy for 1500 test sets. We also investigated the influences of the defect-tolerance capability of the CA, impact of the conductance range of the integrated memristors, and presence or absence of selection functionality in the integrated memristors on the image classification tasks. We offer valuable insights into the behavior and performance of CA devices under various conditions and provide evidence of the practicality of memristor-integrated passive CAs as hardware accelerators for NN applications.https://doi.org/10.1038/s41467-023-44620-1 |
| spellingShingle | Kanghyeok Jeon Jin Joo Ryu Seongil Im Hyun Kyu Seo Taeyong Eom Hyunsu Ju Min Kyu Yang Doo Seok Jeong Gun Hwan Kim Purely self-rectifying memristor-based passive crossbar array for artificial neural network accelerators Nature Communications |
| title | Purely self-rectifying memristor-based passive crossbar array for artificial neural network accelerators |
| title_full | Purely self-rectifying memristor-based passive crossbar array for artificial neural network accelerators |
| title_fullStr | Purely self-rectifying memristor-based passive crossbar array for artificial neural network accelerators |
| title_full_unstemmed | Purely self-rectifying memristor-based passive crossbar array for artificial neural network accelerators |
| title_short | Purely self-rectifying memristor-based passive crossbar array for artificial neural network accelerators |
| title_sort | purely self rectifying memristor based passive crossbar array for artificial neural network accelerators |
| url | https://doi.org/10.1038/s41467-023-44620-1 |
| work_keys_str_mv | AT kanghyeokjeon purelyselfrectifyingmemristorbasedpassivecrossbararrayforartificialneuralnetworkaccelerators AT jinjooryu purelyselfrectifyingmemristorbasedpassivecrossbararrayforartificialneuralnetworkaccelerators AT seongilim purelyselfrectifyingmemristorbasedpassivecrossbararrayforartificialneuralnetworkaccelerators AT hyunkyuseo purelyselfrectifyingmemristorbasedpassivecrossbararrayforartificialneuralnetworkaccelerators AT taeyongeom purelyselfrectifyingmemristorbasedpassivecrossbararrayforartificialneuralnetworkaccelerators AT hyunsuju purelyselfrectifyingmemristorbasedpassivecrossbararrayforartificialneuralnetworkaccelerators AT minkyuyang purelyselfrectifyingmemristorbasedpassivecrossbararrayforartificialneuralnetworkaccelerators AT dooseokjeong purelyselfrectifyingmemristorbasedpassivecrossbararrayforartificialneuralnetworkaccelerators AT gunhwankim purelyselfrectifyingmemristorbasedpassivecrossbararrayforartificialneuralnetworkaccelerators |