Abstract
This article presents an area- and power-efficient hardware architecture for the brain-implantable spiking neural networks (SNNs). The proposed generalized hardware architecture is parameterizable and reconfigurable such that the maximum supported number of neurons, the interconnection structure among neurons, and the resolution of the time step can be readily adjusted for realizing various SNN topologies. The designed SNN hardware architecture is capable of emulating moderately-sized SNNs with tens of thousands of neurons in real-time with varying degrees of parallelism, while reducing the resource utilization by 34% for similarly sized SNNs implemented on a single field-programmable gate array (FPGA). We evaluate the model using the MNIST digit recognition benchmark and show that the network can accurately classify handwritten digits with 89.8% accuracy. Compared to the other recently implemented SNN emulators based on FPGAs, the designed and implemented single-FPGA system is able to emulate moderately-sized SNNs instead of using a cluster of FPGAs or CPUs. The application-specific integrated circuit (ASIC) implementation of a moderately-sized SNN is estimated to occupy 3.6 mm2 of silicon area. Post-layout synthesis and simulation results show that the ASIC will dissipate 3.6 mW of power from a 1.16 V supply while operating at 34.7 MHz in a standard 32-nm CMOS process.
Similar content being viewed by others
Availability of data and material
The MNIST datasets used during this study are available at http://yann.lecun.com/exdb/mnist/.
References
National Spinal Cord Injury Statistical Center, Facts and figures at a glance, f Alabama at Birmingham, (2016)
Yin X-X, Sun L, Fu Y, Lu R, Zhang Y (2022) U-net-based medical image segmentation. J Healthcare Eng 15:2022
Lea C, Vidal R, Reiter A, Hager GD (2016) Temporal convolutional networks: a unified approach to action segmentation, in : European Conference on Computer vision
Vaswani A, Shazeer N, Parmar N, Uszkoreit J, Jones L, Gomez AN , Kaiserl, Polosukhin I (2017) Attention is all you need, Adv Neural Inform Process Syst, 30
Raganato A, Tiedemann J (2018) An analysis of encoder representations in transformer-based machine translation, in: EMNLP Workshop BlackboxNLP: analyzing and interpreting neural networks for NLP
Maass W (1997) Networks of spiking neurons: the third generation of neural network models. Neural Netw 10(9):1659–1671
Bonifazi P et al (2013) In vitro large-scale experimental and theoretical studies for the realization of bi-directional brain-prostheses. Front Neural Circuit 7:40
Nawrot A, Pistohl T, Schrader S, Hehl U, Rodriguez V, Aertsen A (2003) Embedding living neurons into simulated neural networks, in: IEEE Conference on Neural Engineering, pp. 229–232
Zbrzeski A et al (2016) Bio-inspired controller on an FPGA applied to closed-loop diaphragmatic stimulation. Front Neurosci 10:275
Mosbacher Y, Khoyratee F, Goldin M, Kanner S, Malakai Y, Silva M, Grassia F, Simon YB, Cortes J, Barzilai A et al (2020) Toward neuroprosthetic real-time communication from in silico to biological neuronal network via patterned optogenetic stimulation. Sci Rep 10(1):1–16
Buccelli S, Bornat Y, Colombi I, Ambroise M, Martines L, Pasquale V, Bisio M, Tessadori J, Nowak P, Grassia F et al (2019) A neuromorphic prosthesis to restore communication in neuronal networks. IScience 19:402–414
Berger T et al (2011) A cortical neural prosthesis for restoring and enhancing memory. J Neural Eng 8(4):046089
Berger T et al (2012) A hippocampal cognitive prosthesis: multi-input, multi-output nonlinear modeling and vlsi implementation. IEEE Transact Neural Syst Rehabilitat Eng 20(2):198–211
Abbott LF (1999) Lapicque’s introduction of the integrate-and-fire model neuron. Brain Res Bullet 50(5):303–304
Hodgkin AL, Huxley AF (1952) A quantitative description of membrane current and its application to conduction and excitation in nerve. J Physiol 117(4):500–544
Izhikevich EM (2003) Simple model of spiking neurons. IEEE Transact Neural Netw 14(6):1569–1572
Izhikevich EM (2004) Which model to use for cortical spiking neurons? IEEE Transact Neural Netw 15(5):1063–1070
Luk W, Thomas D (2009)FPGA accelerated simulation of biologically plausible spiking neural networks, in: IEEE Symposium on field programmable custom computing machines, pp. 45–52
Ambroise M, Levi T, Bornat Y, Saighi S (2013) Biorealistic spiking neural network on FPGA, in: IEEE information sciences and systems, pp. 1–6
Pani D et al (2017) An FPGA platform for real-time simulation of spiking neuronal networks. Front Neurosci 11(90):1–13
Cheung K, Schultz SR, Luk W (2016) NeuroFlow: a general purpose spiking neural network simulation platform using customizable processors. Front Neurosci 9(516):1–15
Khoyratee F, Grassia F, Saïghi S, Levi T (2019) Optimized real-time biomimetic neural network on FPGA for bio-hybridization. Front Neurosci 13:377
Akbarzadeh-Sherbaf K, Abdoli B, Safari S, Vahabie A-H (2018) A scalable FPGA architecture for randomly connected networks of hodgkin-huxley neurons. Front Neurosci 12:698
Furber SB, Galluppi F, Temple S, Plana LA (2014) The spinnaker project. Proceed IEEE 102(5):652–665
Moradi S, Qiao N, Stefanini F, Indiveri G (2017) A scalable multicore architecture with heterogeneous memory structures for dynamic neuromorphic asynchronous processors (DYNAPs). IEEE Transact Biomed Circuits Syst 12(1):106–122
Qiao N, Mostafa H, Corradi F, Osswald M, Stefanini F, Sumislawska D, Indiveri G (2015) A reconfigurable on-line learning spiking neuromorphic processor comprising 256 neurons and 128k synapses. Front Neurosci 9:141
Seo JS, et al. (2011) A 45 nm CMOS neuromorphic chip with a scalable architecture for learning in networks of spiking neurons, in: IEEE Custom Integrated Circuits Conference
Frenkel C et al (2019) A 0.086-mm\(^2\) 12.7-pJ/SOP 64k-synapse 256-neuron online-learning digital spiking neuromorphic processor in 28-nm cmos. IEEE Transact Biomed Circuits Syst 13(1):145–158
Akopyan F et al (2015) Truenorth: Design and tool flow of a 65 mw 1 million neuron programmable neurosynaptic chip. IEEE Transact Comput-Aided Des Integrat Circuits Syst 34(10):1537–1557
Davies M et al (2018) Loihi: A neuromorphic manycore processor with on-chip learning. IEEE Micro 38(1):82–99
Pei J, Deng L, Song S, Zhao M, Zhang Y, Wu S, Wang G, Zou Z, Wu Z, He W et al (2019) Towards artificial general intelligence with hybrid tianjic chip architecture. Nature 572(7767):106–111
Wells RB (2005) “Cortical neurons and circuits: a tutorial introduction [online]. [cit. 2008-02-05],” URL: http://www.mrc.uidaho.edu/\(^{\sim }\)rwells/techdocs/CorticalNeuronsandCircuits.pdfd
Parhi KK, Wang C-Y, Brown AP (1992) Synthesis of control circuits in folded pipelined dsp architectures. IEEE J Solid-State Circuit 27(1):29–43
Kocaturk M, Gulcur HO, Canbeyli R (2015) Toward building hybrid biological/in silico neural networks for motor neuroprosthetic control. Front Neurorobot 9:8
Tsai D, Sawyer D, Bradd A, Yuste R, Shepard KL (2017) A very large-scale microelectrode array for cellular-resolution electrophysiology. Nat Commun 8(1):1802
LeCun Y, Bottou L, Bengio Y, Haffner P (1998) Gradient-based learning applied to document recognition. Proceed IEEE 86(11):2278–2324
Diehl PU, Cook M (2015) Unsupervised learning of digit recognition using spike-timing-dependent plasticity. Front Comput Neurosci 9:99
Tavanaei A, Maida A (2019) Bp-stdp: Approximating backpropagation using spike timing dependent plasticity. Neurocomputing 330:39–47
Neftci E, Das S, Pedroni B, Kreutz-Delgado K, Cauwenberghs G (2014) Event-driven contrastive divergence for spiking neuromorphic systems. Front Neurosci 7:272
Zhao B, Ding R, Chen S, Linares-Barranco B, Tang H (2014) Feedforward categorization on aer motion events using cortex-like features in a spiking neural network. IEEE Transact Neural Netw Learn Syst 26(9):1963–1978
Beyeler M, Dutt ND, Krichmar JL (2013) Categorization and decision-making in a neurobiologically plausible spiking network using a stdp-like learning rule. Neural Netw 48:109–124
Brader JM, Senn W, Fusi S (2007) Learning real-world stimuli in a neural network with spike-driven synaptic dynamics. Neural Comput 19(11):2881–2912
An S, Lee M, Park S, Yang H, So J (2020) “An ensemble of simple convolutional neural network models for mnist digit recognition,” arXiv preprint arXiv:2008.10400
Davison AP, Brüderle D, Eppler JM, Kremkow J, Muller E, Pecevski D, Perrinet L, Yger P (2009) Pynn: a common interface for neuronal network simulators. Front Neuroinform 2:11
Stewart TC, Tripp B, Eliasmith C (2009) Python scripting in the nengo simulator. Front Neuroinform 3:7
Hofstötter C, et al. (2005) The cerebellum chip: an analog VLSI implementation of a cerebellar model of classical conditioning, in: Advances in neural information processing systems, pp. 577–584
Park J, Ha S, Yu T, Neftci E, Cauwenberghs G (2014) A 65k-neuron 73-mevents/s 22-pj/event asynchronous micro-pipelined integrate-and-fire array transceiver,in: IEEE BioCAS Proceedings, pp. 675–678
Mead C (1989) Analog VLSI and neutral systems, NASA STI/Recon Tech Rep A, 90
Joubert A, Belhadj B, Temam O, Héliot R (2012) Hardware spiking neurons design: Analog or digital? in: IEEE International Joint Conference on Neural Networks, pp. 1–5
Stillmaker A, Xiao Z, Baas B (2011)Toward more accurate scaling estimates of cmos circuits from 180 nm to 22 nm,VLSI Computation Lab, ECE Department, University of California, Davis, Tech. Rep. ECE-VCL-2011-4, 4, p. m8
Brette R, Gerstner W (2005) Adaptive exponential integrate-and-fire model as an effective description of neuronal activity. J Neurophysiol 94(5):3637–3642
Naud R, Marcille N, Clopath C, Gerstner W (2008) Firing patterns in the adaptive exponential integrate-and-fire model. Biol Cybernet 99(4–5):335
Pal S, Gupta V, Ki WH, Islam A (2019) Design and development of memristor-based rram. IET Circuits, Dev Syst 13(4):548–557
Chu M, Kim B, Park S, Hwang H, Jeon M, Lee BH, Lee B-G (2014) Neuromorphic hardware system for visual pattern recognition with memristor array and cmos neuron. IEEE Transact Ind Electron 62(4):2410–2419
Shukla A, Ganguly U (2018) An on-chip trainable and the clock-less spiking neural network with 1r memristive synapses. IEEE Transact Biomed Circuits Syst 12(4):884–893
Chen B, Yang H, Zhuge F, Li Y, Chang T-C, He Y-H, Yang W, Xu N, Miao X-S (2019) Optimal tuning of memristor conductance variation in spiking neural networks for online unsupervised learning. IEEE Transact Electron Dev 66(6):2844–2849
Zheng N, Mazumder P (2018) Learning in memristor crossbar-based spiking neural networks through modulation of weight-dependent spike-timing-dependent plasticity. IEEE Transact Nanotechnol 17(3):520–532
Acknowledgements
This work was supported by the Center for Neurotechnology (CNT), a National Science Foundation (NSF) Engineering Research Center (EEC-1028725) and the NSF Award #2007131.
Author information
Authors and Affiliations
Corresponding author
Ethics declarations
Conflict of interest
The authors have no competing interests to declare.
Additional information
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Rights and permissions
Springer Nature or its licensor (e.g. a society or other partner) holds exclusive rights to this article under a publishing agreement with the author(s) or other rightsholder(s); author self-archiving of the accepted manuscript version of this article is solely governed by the terms of such publishing agreement and applicable law.
About this article
Cite this article
Valencia, D., Alimohammad, A. A generalized hardware architecture for real-time spiking neural networks. Neural Comput & Applic 35, 17821–17835 (2023). https://doi.org/10.1007/s00521-023-08650-6
Received:
Accepted:
Published:
Issue Date:
DOI: https://doi.org/10.1007/s00521-023-08650-6