Mapping Spiking Neural Networks to Neuromorphic Hardware
Open accessThe DOI matches its Crossref record; 58 of the 58 references it cites were found in a registry or an index.
Abstract
Neuromorphic hardware implements biological neurons and synapses to execute a spiking neural network (SNN)-based machine learning. We present SpiNeMap, a design methodology to map SNNs to crossbar-based neuromorphic hardware, minimizing spike latency and energy consumption. SpiNeMap operates in two steps: SpiNeCluster and SpiNePlacer. SpiNeCluster is a heuristic-based clustering technique to partition an SNN into clusters of synapses, where intracluster local synapses are mapped within crossbars of the hardware and intercluster global synapses are mapped to the shared interconnect. SpiNeCluster minimizes the number of spikes on global synapses, which reduces spike congestion and improves application performance. SpiNePlacer then finds the best placement of local and global synapses on the hardware using a metaheuristic-based approach to minimize energy consumption and spike latency. We evaluate SpiNeMap using synthetic and realistic SNNs on a state-of-the-art neuromorphic hardware. We show that SpiNeMap reduces average energy consumption by 45% and spike latency by 21%, compared to the best-performing SNN mapping technique.
Overview
The claims and statistics of this paper haven’t been checked yet.
Free: it reads the open-access text and checks the first 5 claims that cite a source we can check. Create a free account to check every claim.
Integrity signals
Registry
Cites
Statistics
Verification
Claims the paper makes
The claims and statistics of this paper haven’t been checked yet.
Free: it reads the open-access text and checks the first 5 claims that cite a source we can check. Create a free account to check every claim.
Self-check
Percentages, sums and reported differences re-computed from the paper’s own numbers.
The self-check needs the paper’s full text.
Citations
Cites
LeCun, Y., Bengio, Y., & Hinton, G. (2015) · Nature · doi:10.1038/nature14539 · Cited by 84,126
Rethinking the Inception Architecture for Computer Vision
Szegedy, C., Vanhoucke, V., Ioffe, S., Shlens, J., & Wojna, Z. (2016) · 2016 IEEE Conference on Computer Vision and Pattern Recognition (CVPR) · doi:10.1109/cvpr.2016.308 · Cited by 32,138
Human-level control through deep reinforcement learning
Mnih, V., Kavukcuoglu, K., Silver, D., Rusu, A. A., Veness, J., Bellemare, M. G., Graves, A., Riedmiller, M., Fidjeland, A. K., Ostrovski, G., et al. (2015) · Nature · doi:10.1038/nature14236 · Cited by 30,357 · cited 1×
“Examples include the liquid state machine [27], zero-shot learning [61], one-shot learning [62], lifelong learning [63], transfer learning [64], and deep reinforcement learning [65] among others.”
BackgroundPan, S. J., & Yang, Q. (2010) · IEEE Transactions on Knowledge and Data Engineering · doi:10.1109/tkde.2009.191 · Cited by 23,387 · cited 1×
“Examples include the liquid state machine [27], zero-shot learning [61], one-shot learning [62], lifelong learning [63], transfer learning [64], and deep reinforcement learning [65] among others.”
BackgroundA new optimizer using particle swarm theory
Eberhart, R., & Kennedy, J. (2002) · MHS'95. Proceedings of the Sixth International Symposium on Micro Machine and Human Science · doi:10.1109/mhs.1995.494215 · Cited by 16,067 · cited 2×
“We use an instance of PSO [28] to find the optimum mapping.”
MethodSpeech recognition with deep recurrent neural networks
Graves, A., Mohamed, A.-R., & Hinton, G. (2013) · 2013 IEEE International Conference on Acoustics, Speech and Signal Processing · doi:10.1109/icassp.2013.6638947 · Cited by 8,893 · cited 1×
“NEUROMORPHIC hardware, such as TrueNorth [1], Loihi [2], and DYNAP-SE [3], can implement machine learning tasks [4]–[6] using spiking neural networks (SNNs) [7]–[9].”
Method(2019) · Handbook of Machine Learning · doi:10.1142/9789811205675_0004 · Cited by 6,270
An Efficient Heuristic Procedure for Partitioning Graphs
Kernighan, B. W., & Lin, S. (1970) · Bell System Technical Journal · doi:10.1002/j.1538-7305.1970.tb01770.x · Cited by 5,305 · cited 2×
“Transforming G(N ,S) → H(C, E) is a classical graph partitioning problem [29] and has been applied in many contexts, including task mapping on multiprocessor systems [30].”
BackgroundMethodMaass, W., Natschläger, T., & Markram, H. (2002) · Neural Computation · doi:10.1162/089976602760407955 · Cited by 4,025 · cited 2×
“Examples include the liquid state machine [27], zero-shot learning [61], one-shot learning [62], lifelong learning [63], transfer learning [64], and deep reinforcement learning [65] among others.”
BackgroundLoihi: A Neuromorphic Manycore Processor with On-Chip Learning
Davies, M., Srinivasa, N., Lin, T.-H., Chinya, G., Cao, Y., Choday, S. H., Dimou, G., Joshi, P., Imam, N., Jain, S., et al. (2018) · IEEE Micro · doi:10.1109/mm.2018.112130359 · Cited by 3,897
Networks of spiking neurons: The third generation of neural network models
Maass, W. (1997) · Neural Networks · doi:10.1016/s0893-6080(97)00011-7 · Cited by 3,481 · cited 1×
“NEUROMORPHIC hardware, such as TrueNorth [1], Loihi [2], and DYNAP-SE [3], can implement machine learning tasks [4]–[6] using spiking neural networks (SNNs) [7]–[9].”
MethodStriving for Simplicity: The All Convolutional Net
Springenberg, J. T., Dosovitskiy, A., Brox, T., & Riedmiller, M. (2014) · arXiv · doi:10.48550/arxiv.1412.6806 · Cited by 3,447 · cited 2×
“We use eight realistic applications: image smoothing (ImgSmooth) [28] on 64x64 images, edge detection (EdgeDet) [28] on 64x64 images using difference-of-Gaussian, multi-layer perceptron (MLP)-based handwritten digit recognition (MLP-MNIST) [41] on 28x28 images of handwritten digits, ECG-based heart-rate estimation (He…”
MethodEyeriss: An Energy-Efficient Reconfigurable Accelerator for Deep Convolutional Neural Networks
Chen, Y.-H., Krishna, T., Emer, J. S., & Sze, V. (2017) · IEEE Journal of Solid-State Circuits · doi:10.1109/jssc.2016.2616357 · Cited by 3,247
One-shot learning of object categories
Fei-Fei, L., Fergus, R., & Perona, P. (2006) · IEEE Transactions on Pattern Analysis and Machine Intelligence · doi:10.1109/tpami.2006.79 · Cited by 3,029 · cited 1×
“Examples include the liquid state machine [27], zero-shot learning [61], one-shot learning [62], lifelong learning [63], transfer learning [64], and deep reinforcement learning [65] among others.”
BackgroundA Linear-Time Heuristic for Improving Network Partitions
Fiduccia, C. M., & Mattheyses, R. M. (1982) · ACM/IEEE Design Automation Conference (DAC) · doi:10.1109/dac.1982.1585498 · Cited by 2,567
Memristive crossbar arrays for brain-inspired computing
Xia, Q., & Yang, J. J. (2019) · Nature Materials · doi:10.1038/s41563-019-0291-x · Cited by 1,800
Unsupervised learning of digit recognition using spike-timing-dependent plasticity
Diehl, P. U., & Cook, M. (2015) · Frontiers in Computational Neuroscience · doi:10.3389/fncom.2015.00099 · Cited by 1,452 · cited 4×
“We use eight realistic applications: image smoothing (ImgSmooth) [28] on 64x64 images, edge detection (EdgeDet) [28] on 64x64 images using difference-of-Gaussian, multi-layer perceptron (MLP)-based handwritten digit recognition (MLP-MNIST) [41] on 28x28 images of handwritten digits, ECG-based heart-rate estimation (He…”
BackgroundMethodInfluentialNeuromorphic computing using non-volatile memory
Burr, G. W., Shelby, R. M., Sebastian, A., Kim, S., Kim, S., Sidler, S., Virwani, K., Ishii, M., Narayanan, P., Fumarola, A., et al. (2017) · Advances in Physics: X · doi:10.1080/23746149.2016.1259585 · Cited by 1,107
Spike Timing-Dependent Plasticity of Neural Circuits
Dan, Y., & Poo, M.-M. (2004) · Neuron · doi:10.1016/j.neuron.2004.09.007 · Cited by 1,040
Gewaltig, M.-O., & Diesmann, M. (2007) · Scholarpedia · doi:10.4249/scholarpedia.1430 · Cited by 1,025
Zero-Shot Learning Through Cross-Modal Transfer
Socher, R., Ganjoo, M., Sridhar, H., Bastani, O., Manning, C. D., & Ng, A. Y. (2013) · arXiv · doi:10.48550/arxiv.1301.3666 · Cited by 997 · cited 1×
“Emerging machine learning approaches, such as [61]–[66], can also be mapped to the neuromorphic”
BackgroundSpiking Deep Convolutional Neural Networks for Energy-Efficient Object Recognition
Cao, Y., Chen, Y., & Khosla, D. (2015) · International Journal of Computer Vision · doi:10.1007/s11263-014-0788-3 · Cited by 917
New Generation of Predictive Technology Model for Sub-45 nm Early Design Exploration
Zhao, W., & Cao, Y. (2006) · IEEE Transactions on Electron Devices · doi:10.1109/ted.2006.884077 · Cited by 852 · cited 1×
“Technology parameters are obtained from [37] for 45-nm technology node [38].”
A detailed and flexible cycle-accurate Network-on-Chip simulator
Jiang, N., Balfour, J., Becker, D. U., Towles, B., Dally, W. J., Michelogiannakis, G., & Kim, J. (2013) · 2013 IEEE International Symposium on Performance Analysis of Systems and Software (ISPASS) · doi:10.1109/ispass.2013.6557149 · Cited by 790 · cited 1×
“SpiNeMap can be used with other interconnect simulators, such as [67]–[69], which also supports cycle-accurate and trace-driven simulation.”
MethodMoradi, S., Qiao, N., Stefanini, F., & Indiveri, G. (2018) · IEEE Transactions on Biomedical Circuits and Systems · doi:10.1109/tbcas.2017.2759700 · Cited by 650 · cited 5×
“NEUROMORPHIC hardware, such as TrueNorth [1], Loihi [2], and DYNAP-SE [3], can implement machine learning tasks [4]–[6] using spiking neural networks (SNNs) [7]–[9].”
MethodInfluentialSelf-citationSome simplified NP-complete problems
Garey, M. R., Johnson, D. S., & Stockmeyer, L. (1974) · Proceedings of the sixth annual ACM symposium on Theory of computing - STOC '74 · doi:10.1145/800119.803884 · Cited by 614
PyNN: a common interface for neuronal network simulators
Davison, A. P. (2008) · Frontiers in Neuroinformatics · doi:10.3389/neuro.11.011.2008 · Cited by 607
Goodman, D. F. M. (2009) · Frontiers in Neuroscience · doi:10.3389/neuro.01.026.2009 · Cited by 433
Mapping on multi/many-core systems: survey of current and emerging trends
Singh, A. K., Shafique, M., Kumar, A. S., & Henkel, J. (2013) · ACM/IEEE Design Automation Conference (DAC) · doi:10.1145/2463209.2488734 · Cited by 428
(2009) · InTech eBooks · doi:10.5772/109 · Cited by 295
(2013) · Cited by 290
Philosophy of the Spike: Rate-Based vs. Spike-Based Theories of the Brain
Brette, R. (2015) · Frontiers in Systems Neuroscience · doi:10.3389/fnsys.2015.00151 · Cited by 270 · cited 1×
“Compared to analog and rate models, machine learning techniques implemented with spike model [49] and brain-inspired learning algorithms [50], e.g., spiking neural networks [1], have ultra-low power footprint when executed on neuromorphic hardware such as DYNAP-SE [7], TrueNorth [ ?”
MethodSpiNNaker: Mapping neural networks onto a massively-parallel chip multiprocessor
Khan, M. M., Lester, D. R., Plana, L. A., Rast, A., Jin, X., Painkras, E., & Furber, S. B. (2008) · 2008 IEEE International Joint Conference on Neural Networks (IEEE World Congress on Computational Intelligence) · doi:10.1109/ijcnn.2008.4634199 · Cited by 270 · cited 1×
“propose a mapping strategy for SNNs on the SpiNNaker platform [55].”
MethodTrueNorth: Accelerating From Zero to 64 Million Neurons in 10 Years
DeBole, M. V., Taba, B., Amir, A., Akopyan, F., Andreopoulos, A., Risk, W. P., Kusnitz, J., Ortega Otero, C., Nayak, T. K., Appuswamy, R., et al. (2019) · Computer · doi:10.1109/mc.2019.2903009 · Cited by 232 · cited 8×
“Examples include the multi-stage networks-on-chip for the new TrueNorth platform [20] and the segmented bus for the new DynapSE platform [21].”
BackgroundMethodInfluentialLee, H. G., Chang, N., Ogras, U. Y., & Marculescu, R. (2007) · ACM Transactions on Design Automation of Electronic Systems · doi:10.1145/1255456.1255460 · Cited by 231
Spike-Timing-Dependent Hebbian Plasticity as Temporal Difference Learning
Rao, R. P. N., & Sejnowski, T. J. (2001) · Neural Computation · doi:10.1162/089976601750541787 · Cited by 226 · cited 1×
“Beyond this supervised approach, recent works [22] have also developed peripheral structures necessary to implement online synaptic updates such as spike timing dependent plasticity (STDP) [23].”
BackgroundAnalysis of Parallel Spike Trains
(2010) · Springer eBooks · doi:10.1007/978-1-4419-5675-0 · Cited by 206
Diehl, P. U., Zarrella, G., Cassidy, A., Pedroni, B. U., & Neftci, E. (2016) · 2016 IEEE International Conference on Rebooting Computing (ICRC) · doi:10.1109/icrc.2016.7738691 · Cited by 202
GeNN: a code generation framework for accelerated brain simulations
Yavuz, E., Turner, J., & Nowotny, T. (2016) · Scientific Reports · doi:10.1038/srep18854 · Cited by 169
Neuromorphic architectures for spiking deep neural networks
Indiveri, G., Corradi, F., & Qiao, N. (2015) · 2015 IEEE International Electron Devices Meeting (IEDM) · doi:10.1109/iedm.2015.7409623 · Cited by 163 · cited 1×
“Technology parameters are obtained from [37] for 45-nm technology node [38].”
Self-citationNeuromorphic Computing with Memristor Crossbar
Zhang, X., Huang, A., Hu, Q., Xiao, Z., & Chu, P. K.-H. (2018) · physica status solidi (a) · doi:10.1002/pssa.201700875 · Cited by 138
Reservoir-based techniques for speech recognition
Verstraeten, D., Schrauwen, B., & Stroobandt, D. (2006) · The 2006 IEEE International Joint Conference on Neural Network Proceedings · doi:10.1109/ijcnn.2006.1716215 · Cited by 131
Wijesinghe, P., Ankit, A., Sengupta, A., & Roy, K. (2018) · IEEE Transactions on Emerging Topics in Computational Intelligence · doi:10.1109/tetci.2018.2829924 · Cited by 125 · cited 2×
“We make this choice because by utilizing the minimum number of crossbars, the overall energy consumption of the hardware can be minimized [10]–[14].”
MethodNEUTRAMS: Neural network transformation and co-design under neuromorphic hardware constraints
Ji, Y., Zhang, Y., Li, S., Chi, P., Jiang, C., Qu, P., Xie, Y., & Chen, W. (2016) · IEEE/ACM International Symposium on Microarchitecture (MICRO) · doi:10.1109/micro.2016.7783724 · Cited by 121 · cited 4×
“1) The baseline [19] minimizes the use of crossbars.”
BackgroundMethodInfluentialChou, T.-S., Kashyap, H. J., Xing, J., Listopad, S., Rounds, E. L., Beyeler, M., Dutt, N. D., & Krichmar, J. L. (2018) · International Joint Conference on Neural Networks (IJCNN) · doi:10.1109/ijcnn.2018.8489326 · Cited by 120 · cited 4×
“The realistic applications are image smoothing (ImgSmooth) [27] on 64 × 64 images, edge detection (EdgeDet) [27] on 64 × 64 images using the difference-of-Gaussian, multilayer perceptron (MLP)-based handwritten digit recognition (MLP-MNIST) [9] on 28 × 28 images of handwritten digits, ECG-based heart-rate estimation (…”
MethodInfluentialSelf-citationA hierachical configuration system for a massively parallel neural hardware platform
Galluppi, F., Davies, S., Rast, A. D., Sharp, T., Plana, L. A., & Furber, S. B. (2012) · ACM International Conference on Computing Frontiers (CF) · doi:10.1145/2212908.2212934 · Cited by 97 · cited 7×
“These include the PACMAN [16], NEUTRAMS [17], and PSOPART [18].”
BackgroundMethodInfluentialUnsupervised heart-rate estimation in wearables with Liquid states and a probabilistic readout
Das, A. K., Pradhapan, P., Groenendaal, W., Adiraju, P., Rajan, R. T., Catthoor, F., Schaafsma, S., Krichmar, J. L., Dutt, N. D., & Hoof, C. A. V. (2018) · Neural Networks · doi:10.1016/j.neunet.2017.12.015 · Cited by 95 · cited 2×
“The realistic applications are image smoothing (ImgSmooth) [27] on 64 × 64 images, edge detection (EdgeDet) [27] on 64 × 64 images using the difference-of-Gaussian, multilayer perceptron (MLP)-based handwritten digit recognition (MLP-MNIST) [9] on 28 × 28 images of handwritten digits, ECG-based heart-rate estimation (…”
BackgroundMethodSelf-citationSPINDLE: SPINtronic deep learning engine for large-scale neuromorphic computing
Ramasubramanian, S. G., Venkatesan, R., Sharad, M., Roy, K., & Raghunathan, A. (2014) · Proceedings of the 2014 international symposium on Low power electronics and design · doi:10.1145/2627369.2627625 · Cited by 85 · cited 1×
“[48] use spin-transfer torque magnetic RAM (STT MRAM), Burr et al.”
MethodA System-Level Simulator for RRAM-Based Neuromorphic Computing Chips
Lee, M. K. F., Cui, Y., Somu, T., Luo, T., Zhou, J. R., Tang, W. T., Wong, W., & Goh, R. S. M. (2018) · ACM Transactions on Architecture and Code Optimization · doi:10.1145/3291054 · Cited by 69 · cited 15×
“Our SpiNeMap framework is open-sourced and can be downloaded from the url https://github.com/drexel-DISCO/SpiNeMap .”
InfluentialInterspike interval embedding of chaotic signals
SAUER, T. D. (1995) · Chaos An Interdisciplinary Journal of Nonlinear Science · doi:10.1063/1.166094 · Cited by 68 · cited 1×
“Unfortunately, a shared interconnect introduces latency, which distorts interspike intervals (ISIs) [11].”
BackgroundMapping of local and global synapses on spiking neuromorphic hardware
Das, A., Wu, Y., Huynh, K., Dell'Anna, F., Catthoor, F., & Schaafsma, S. (2018) · Design, Automation, and Test in Europe · doi:10.23919/date.2018.8342201 · Cited by 64 · cited 6×
“2(b) illustrates PSOPART [20], which minimizes the number of spikes on the shared interconnect in step 2 using an instance of the particle swarm optimization (PSO) [28].”
BackgroundMethodInfluentialSelf-citationNEUTRAMS: neural network transformation and co-design under neuromorphic hardware constraints
Ji, Y., Zhang, Y., Li, S., Chi, P., Jiang, C., Qu, P., Xie, Y., & Chen, W. (2016) · IEEE/ACM International Symposium on Microarchitecture (MICRO) · doi:10.5555/3195638.3195663 · Cited by 57
Power-Accuracy Trade-Offs for Heartbeat Classification on Neural Networks Hardware
Balaji, A., Corradi, F., Das, A. K., Pande, S., Schaafsma, S., & Catthoor, F. (2018) · Journal of Low Power Electronics · doi:10.1166/jolpe.2018.1582 · Cited by 46 · cited 1×
“The realistic applications are image smoothing (ImgSmooth) [27] on 64 × 64 images, edge detection (EdgeDet) [27] on 64 × 64 images using the difference-of-Gaussian, multilayer perceptron (MLP)-based handwritten digit recognition (MLP-MNIST) [9] on 28 × 28 images of handwritten digits, ECG-based heart-rate estimation (…”
MethodSelf-citationAn EDA framework for large scale hybrid neuromorphic computing systems
Wen, W., Wu, C.-R., Hu, X., Liu, B., Ho, T.-Y., Li, X., & Chen, Y. (2015) · ACM/IEEE Design Automation Conference (DAC) · doi:10.1145/2744769.2744795 · Cited by 42 · cited 1×
“Recent works, such as [12]–[17], uses a single large crossbar to map SNNs.”
BackgroundDesign-technology co-optimization for OxRRAM-based synaptic processing unit
Mallik, A., Garbin, D., Fantini, A., Rodopoulos, D., Degraeve, R., Stuijt, J., Das, A. K., Schaafsma, S., Debacker, P., Donadio, G. L., et al. (2017) · Symposium on VLSI Technology · doi:10.23919/vlsit.2017.7998166 · Cited by 40 · cited 1×
“propose to use oxidebased resistive RAM (OxRAM) as alternative [54].”
MethodSelf-citationBichler, O., Roclin, D., Gamrat, C., & Querlioz, D. (2013) · IEEE/ACM International Symposium on Nanoscale Architectures (NANOARCH) · doi:10.1109/nanoarch.2013.6623029 · Cited by 30 · cited 1×
“SpiNeMap can be combined with any other SNN simulators [52]–[56].”
BackgroundFacial expression recognition based on Liquid State Machines built of alternative neuron models
Grzyb, B. J., Chinellato, E., Wójcik, G. M., & Kamiński, W. A. (2009) · International Joint Conference on Neural Networks (IJCNN) · doi:10.1109/ijcnn.2009.5179025 · Cited by 26 · cited 1×
“[47] use spiking liquid-state machine for facial recognition.”
MethodNeuromorphic Computing Across the Stack: Devices, Circuits and Architectures
Ankit, A., Sengupta, A., & Roy, K. (2018) · SiPS ... design and implementation - IEEE Workshop on Signal Processing Systems · doi:10.1109/sips.2018.8598419 · Cited by 9 · cited 1×
“Recent works, such as [12]–[17], uses a single large crossbar to map SNNs.”
Background
Cited by
Mysore, N., Hota, G., Deiss, S. R., Pedroni, B. U., & Cauwenberghs, G. (2022) · Frontiers in Neuroscience · doi:10.3389/fnins.2021.797654 · Cited by 7
Metadata
- Authors
- Balaji, A., Catthoor, F., Das, A., Wu, Y., Huynh, K., Dell'Anna, F. G., Indiveri, G., Krichmar, J. L., Dutt, N. D., & Schaafsma, S.
- Venue
- IEEE Transactions on Very Large Scale Integration (VLSI) Systems
- Published
- 2020-01
- Volume
- 28
- Issue
- 1
- Pages
- 76-86
- ISSN
- 1063-8210 (ISSN-L)
- 1557-9999
- OpenAlex
- W2989683650
- Cited by
- 159
- References
- 58
- FWCI
- 6.08
- Type
- journal article
- Language
- en
- Open access
- green
About these checks
Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.
Statistics and arithmetic are recomputed, not judged.
Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.
What counts as a claim ›How we measure confidence ›How we check this ›
Are you an author, or do you see a problem with this page? Request a review
Data sources
Bibliographic data: Semantic Scholar, OpenAlex, Crossref.
Contains information from the Semantic Scholar Academic Graph Datasets, made available under the ODC Attribution License (ODC-BY 1.0).
Semantic Scholar’s data is described in: Kinney, R., et al. (2023). The Semantic Scholar Open Data Platform. arXiv:2301.10140.