MONETA: A Processing-In-Memory-Based Hardware Platform for the Hybrid Convolutional Spiking Neural Network With Online Learning
We present a processing-in-memory (PIM)-based hardware platform, referred to as MONETA, for on-chip acceleration of inference and learning in hybrid convolutional spiking neural network. MONETAuses 8T static random-access memory (SRAM)-based PIM cores for vector matrix multiplication (VMM) augmented...
| Main Authors: | , , , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
Frontiers Media S.A.
2022-04-01
|
| Series: | Frontiers in Neuroscience |
| Subjects: | |
| Online Access: | https://www.frontiersin.org/articles/10.3389/fnins.2022.775457/full |
| _version_ | 1819048217014173696 |
|---|---|
| author | Daehyun Kim Biswadeep Chakraborty Xueyuan She Edward Lee Beomseok Kang Saibal Mukhopadhyay |
| author_facet | Daehyun Kim Biswadeep Chakraborty Xueyuan She Edward Lee Beomseok Kang Saibal Mukhopadhyay |
| author_sort | Daehyun Kim |
| collection | DOAJ |
| description | We present a processing-in-memory (PIM)-based hardware platform, referred to as MONETA, for on-chip acceleration of inference and learning in hybrid convolutional spiking neural network. MONETAuses 8T static random-access memory (SRAM)-based PIM cores for vector matrix multiplication (VMM) augmented with spike-time-dependent-plasticity (STDP) based weight update. The spiking neural network (SNN)-focused data flow is presented to minimize data movement in MONETAwhile ensuring learning accuracy. MONETAsupports on-line and on-chip training on PIM architecture. The STDP-trained convolutional neural network within SNN (ConvSNN) with the proposed data flow, 4-bit input precision, and 8-bit weight precision shows only 1.63% lower accuracy in CIFAR-10 compared to the STDP accuracy implemented by the software. Further, the proposed architecture is used to accelerate a hybrid SNN architecture that couples off-chip supervised (back propagation through time) and on-chip unsupervised (STDP) training. We also evaluate the hybrid network architecture with the proposed data flow. The accuracy of this hybrid network is 10.84% higher than STDP trained accuracy result and 1.4% higher compared to the backpropagated training-based ConvSNN result with the CIFAR-10 dataset. Physical design of MONETAin 65 nm complementary metal-oxide-semiconductor (CMOS) shows 18.69 tera operation per second (TOPS)/W, 7.25 TOPS/W and 10.41 TOPS/W power efficiencies for the inference mode, learning mode, and hybrid learning mode, respectively. |
| first_indexed | 2024-12-21T11:12:44Z |
| format | Article |
| id | doaj.art-1ede96b007f340e3a27859ffc0b1a887 |
| institution | Directory Open Access Journal |
| issn | 1662-453X |
| language | English |
| last_indexed | 2024-12-21T11:12:44Z |
| publishDate | 2022-04-01 |
| publisher | Frontiers Media S.A. |
| record_format | Article |
| series | Frontiers in Neuroscience |
| spelling | doaj.art-1ede96b007f340e3a27859ffc0b1a8872022-12-21T19:06:01ZengFrontiers Media S.A.Frontiers in Neuroscience1662-453X2022-04-011610.3389/fnins.2022.775457775457MONETA: A Processing-In-Memory-Based Hardware Platform for the Hybrid Convolutional Spiking Neural Network With Online LearningDaehyun KimBiswadeep ChakrabortyXueyuan SheEdward LeeBeomseok KangSaibal MukhopadhyayWe present a processing-in-memory (PIM)-based hardware platform, referred to as MONETA, for on-chip acceleration of inference and learning in hybrid convolutional spiking neural network. MONETAuses 8T static random-access memory (SRAM)-based PIM cores for vector matrix multiplication (VMM) augmented with spike-time-dependent-plasticity (STDP) based weight update. The spiking neural network (SNN)-focused data flow is presented to minimize data movement in MONETAwhile ensuring learning accuracy. MONETAsupports on-line and on-chip training on PIM architecture. The STDP-trained convolutional neural network within SNN (ConvSNN) with the proposed data flow, 4-bit input precision, and 8-bit weight precision shows only 1.63% lower accuracy in CIFAR-10 compared to the STDP accuracy implemented by the software. Further, the proposed architecture is used to accelerate a hybrid SNN architecture that couples off-chip supervised (back propagation through time) and on-chip unsupervised (STDP) training. We also evaluate the hybrid network architecture with the proposed data flow. The accuracy of this hybrid network is 10.84% higher than STDP trained accuracy result and 1.4% higher compared to the backpropagated training-based ConvSNN result with the CIFAR-10 dataset. Physical design of MONETAin 65 nm complementary metal-oxide-semiconductor (CMOS) shows 18.69 tera operation per second (TOPS)/W, 7.25 TOPS/W and 10.41 TOPS/W power efficiencies for the inference mode, learning mode, and hybrid learning mode, respectively.https://www.frontiersin.org/articles/10.3389/fnins.2022.775457/fullspiking neural network (SNN)processing-in-memory (PIM)convolutional spiking neural networkon-line learningon-chip learningspike-time-dependent plasticity (STDP) |
| spellingShingle | Daehyun Kim Biswadeep Chakraborty Xueyuan She Edward Lee Beomseok Kang Saibal Mukhopadhyay MONETA: A Processing-In-Memory-Based Hardware Platform for the Hybrid Convolutional Spiking Neural Network With Online Learning Frontiers in Neuroscience spiking neural network (SNN) processing-in-memory (PIM) convolutional spiking neural network on-line learning on-chip learning spike-time-dependent plasticity (STDP) |
| title | MONETA: A Processing-In-Memory-Based Hardware Platform for the Hybrid Convolutional Spiking Neural Network With Online Learning |
| title_full | MONETA: A Processing-In-Memory-Based Hardware Platform for the Hybrid Convolutional Spiking Neural Network With Online Learning |
| title_fullStr | MONETA: A Processing-In-Memory-Based Hardware Platform for the Hybrid Convolutional Spiking Neural Network With Online Learning |
| title_full_unstemmed | MONETA: A Processing-In-Memory-Based Hardware Platform for the Hybrid Convolutional Spiking Neural Network With Online Learning |
| title_short | MONETA: A Processing-In-Memory-Based Hardware Platform for the Hybrid Convolutional Spiking Neural Network With Online Learning |
| title_sort | moneta a processing in memory based hardware platform for the hybrid convolutional spiking neural network with online learning |
| topic | spiking neural network (SNN) processing-in-memory (PIM) convolutional spiking neural network on-line learning on-chip learning spike-time-dependent plasticity (STDP) |
| url | https://www.frontiersin.org/articles/10.3389/fnins.2022.775457/full |
| work_keys_str_mv | AT daehyunkim monetaaprocessinginmemorybasedhardwareplatformforthehybridconvolutionalspikingneuralnetworkwithonlinelearning AT biswadeepchakraborty monetaaprocessinginmemorybasedhardwareplatformforthehybridconvolutionalspikingneuralnetworkwithonlinelearning AT xueyuanshe monetaaprocessinginmemorybasedhardwareplatformforthehybridconvolutionalspikingneuralnetworkwithonlinelearning AT edwardlee monetaaprocessinginmemorybasedhardwareplatformforthehybridconvolutionalspikingneuralnetworkwithonlinelearning AT beomseokkang monetaaprocessinginmemorybasedhardwareplatformforthehybridconvolutionalspikingneuralnetworkwithonlinelearning AT saibalmukhopadhyay monetaaprocessinginmemorybasedhardwareplatformforthehybridconvolutionalspikingneuralnetworkwithonlinelearning |