MB-CNN: Memristive Binary Convolutional Neural Networks for Embedded Mobile Devices
Applications of neural networks have gained significant importance in embedded mobile devices and Internet of Things (IoT) nodes. In particular, convolutional neural networks have emerged as one of the most powerful techniques in computer vision, speech recognition, and AI applications that can impr...
| Main Authors: | , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
MDPI AG
2018-10-01
|
| Series: | Journal of Low Power Electronics and Applications |
| Subjects: | |
| Online Access: | http://www.mdpi.com/2079-9268/8/4/38 |
| _version_ | 1811184060314157056 |
|---|---|
| author | Arjun Pal Chowdhury Pranav Kulkarni Mahdi Nazm Bojnordi |
| author_facet | Arjun Pal Chowdhury Pranav Kulkarni Mahdi Nazm Bojnordi |
| author_sort | Arjun Pal Chowdhury |
| collection | DOAJ |
| description | Applications of neural networks have gained significant importance in embedded mobile devices and Internet of Things (IoT) nodes. In particular, convolutional neural networks have emerged as one of the most powerful techniques in computer vision, speech recognition, and AI applications that can improve the mobile user experience. However, satisfying all power and performance requirements of such low power devices is a significant challenge. Recent work has shown that binarizing a neural network can significantly improve the memory requirements of mobile devices at the cost of minor loss in accuracy. This paper proposes MB-CNN, a memristive accelerator for binary convolutional neural networks that perform XNOR convolution in-situ novel 2R memristive data blocks to improve power, performance, and memory requirements of embedded mobile devices. The proposed accelerator achieves at least 13.26 × , 5.91 × , and 3.18 × improvements in the system energy efficiency (computed by energy × delay) over the state-of-the-art software, GPU, and PIM architectures, respectively. The solution architecture which integrates CPU, GPU and MB-CNN outperforms every other configuration in terms of system energy and execution time. |
| first_indexed | 2024-04-11T13:06:53Z |
| format | Article |
| id | doaj.art-4048cebd0cc844a389d8a76a59d467ce |
| institution | Directory Open Access Journal |
| issn | 2079-9268 |
| language | English |
| last_indexed | 2024-04-11T13:06:53Z |
| publishDate | 2018-10-01 |
| publisher | MDPI AG |
| record_format | Article |
| series | Journal of Low Power Electronics and Applications |
| spelling | doaj.art-4048cebd0cc844a389d8a76a59d467ce2022-12-22T04:22:41ZengMDPI AGJournal of Low Power Electronics and Applications2079-92682018-10-01843810.3390/jlpea8040038jlpea8040038MB-CNN: Memristive Binary Convolutional Neural Networks for Embedded Mobile DevicesArjun Pal Chowdhury0Pranav Kulkarni1Mahdi Nazm Bojnordi2School of Computing, University of Utah, Salt Lake City, UT 84112, USAElectrical & Computer Engineering, University of Utah, Salt Lake City, UT 84112, USASchool of Computing, University of Utah, Salt Lake City, UT 84112, USAApplications of neural networks have gained significant importance in embedded mobile devices and Internet of Things (IoT) nodes. In particular, convolutional neural networks have emerged as one of the most powerful techniques in computer vision, speech recognition, and AI applications that can improve the mobile user experience. However, satisfying all power and performance requirements of such low power devices is a significant challenge. Recent work has shown that binarizing a neural network can significantly improve the memory requirements of mobile devices at the cost of minor loss in accuracy. This paper proposes MB-CNN, a memristive accelerator for binary convolutional neural networks that perform XNOR convolution in-situ novel 2R memristive data blocks to improve power, performance, and memory requirements of embedded mobile devices. The proposed accelerator achieves at least 13.26 × , 5.91 × , and 3.18 × improvements in the system energy efficiency (computed by energy × delay) over the state-of-the-art software, GPU, and PIM architectures, respectively. The solution architecture which integrates CPU, GPU and MB-CNN outperforms every other configuration in terms of system energy and execution time.http://www.mdpi.com/2079-9268/8/4/38convolutional neural networksbinary convolutionsin-situ processingRRAM technologycomputer architectureembedded systems |
| spellingShingle | Arjun Pal Chowdhury Pranav Kulkarni Mahdi Nazm Bojnordi MB-CNN: Memristive Binary Convolutional Neural Networks for Embedded Mobile Devices Journal of Low Power Electronics and Applications convolutional neural networks binary convolutions in-situ processing RRAM technology computer architecture embedded systems |
| title | MB-CNN: Memristive Binary Convolutional Neural Networks for Embedded Mobile Devices |
| title_full | MB-CNN: Memristive Binary Convolutional Neural Networks for Embedded Mobile Devices |
| title_fullStr | MB-CNN: Memristive Binary Convolutional Neural Networks for Embedded Mobile Devices |
| title_full_unstemmed | MB-CNN: Memristive Binary Convolutional Neural Networks for Embedded Mobile Devices |
| title_short | MB-CNN: Memristive Binary Convolutional Neural Networks for Embedded Mobile Devices |
| title_sort | mb cnn memristive binary convolutional neural networks for embedded mobile devices |
| topic | convolutional neural networks binary convolutions in-situ processing RRAM technology computer architecture embedded systems |
| url | http://www.mdpi.com/2079-9268/8/4/38 |
| work_keys_str_mv | AT arjunpalchowdhury mbcnnmemristivebinaryconvolutionalneuralnetworksforembeddedmobiledevices AT pranavkulkarni mbcnnmemristivebinaryconvolutionalneuralnetworksforembeddedmobiledevices AT mahdinazmbojnordi mbcnnmemristivebinaryconvolutionalneuralnetworksforembeddedmobiledevices |