{"status":"ok","message-type":"work","message-version":"1.0.0","message":{"indexed":{"date-parts":[[2025,2,22]],"date-time":"2025-02-22T00:26:04Z","timestamp":1740183964437,"version":"3.37.3"},"reference-count":54,"publisher":"IOP Publishing","issue":"3","license":[{"start":{"date-parts":[[2023,8,31]],"date-time":"2023-08-31T00:00:00Z","timestamp":1693440000000},"content-version":"vor","delay-in-days":0,"URL":"http:\/\/creativecommons.org\/licenses\/by\/4.0"},{"start":{"date-parts":[[2023,8,31]],"date-time":"2023-08-31T00:00:00Z","timestamp":1693440000000},"content-version":"tdm","delay-in-days":0,"URL":"https:\/\/iopscience.iop.org\/info\/page\/text-and-data-mining"}],"funder":[{"name":"PMRF, Govt. of India"}],"content-domain":{"domain":["iopscience.iop.org"],"crossmark-restriction":false},"short-container-title":["Neuromorph. Comput. Eng."],"published-print":{"date-parts":[[2023,9,1]]},"abstract":"<jats:title>Abstract<\/jats:title>\n               <jats:p>Spiking neural networks (SNNs) have emerged as a hardware efficient architecture for classification tasks. The challenge of spike-based encoding has been the lack of a universal training mechanism performed entirely using spikes. There have been several attempts to adopt the powerful backpropagation (BP) technique used in non-spiking artificial neural networks (ANNs): (1) SNNs can be trained by externally computed numerical gradients. (2) A major advancement towards native spike-based learning has been the use of approximate BP using spike-time dependent plasticity with phased forward\/backward passes. However, the transfer of information between such phases for gradient and weight update calculation necessitates external memory and computational access. This is a challenge for standard neuromorphic hardware implementations. In this paper, we propose a stochastic SNN based back-prop (SSNN-BP) algorithm that utilizes a composite neuron to <jats:italic>simultaneously<\/jats:italic> compute the forward pass activations and backward pass gradients explicitly with spikes. Although signed gradient values are a challenge for spike-based representation, we tackle this by splitting the gradient signal into positive and negative streams. The composite neuron encodes information in the form of stochastic spike-trains and converts BP weight updates into temporally and spatially local spike coincidence updates compatible with hardware-friendly resistive processing units. Furthermore, we characterize the quantization effect of discrete spike-based weight update to show that our method approaches BP ANN baseline with sufficiently long spike-trains. Finally, we show that the well-performing softmax cross-entropy loss function can be implemented through inhibitory lateral connections enforcing a winner take all rule. Our SNN with a two-layer network shows excellent generalization through comparable performance to ANNs with equivalent architecture and regularization parameters on static image datasets like MNIST, Fashion-MNIST, Extended MNIST, and temporally encoded image datasets like Neuromorphic MNIST datasets. Thus, SSNN-BP enables BP compatible with purely spike-based neuromorphic hardware.<\/jats:p>","DOI":"10.1088\/2634-4386\/acf1c5","type":"journal-article","created":{"date-parts":[[2023,8,18]],"date-time":"2023-08-18T22:30:41Z","timestamp":1692397841000},"page":"034009","update-policy":"https:\/\/doi.org\/10.1088\/crossmark-policy","source":"Crossref","is-referenced-by-count":0,"title":["A temporally and spatially local spike-based backpropagation algorithm to enable training in hardware"],"prefix":"10.1088","volume":"3","author":[{"ORCID":"https:\/\/orcid.org\/0000-0002-2363-5864","authenticated-orcid":true,"given":"Anmol","family":"Biswas","sequence":"first","affiliation":[]},{"ORCID":"https:\/\/orcid.org\/0000-0001-9191-1632","authenticated-orcid":true,"given":"Vivek","family":"Saraswat","sequence":"additional","affiliation":[]},{"given":"Udayan","family":"Ganguly","sequence":"additional","affiliation":[]}],"member":"266","published-online":{"date-parts":[[2023,8,31]]},"reference":[{"key":"nceacf1c5bib1","doi-asserted-by":"publisher","first-page":"1659","DOI":"10.1016\/S0893-6080(97)00011-7","article-title":"Networks of spiking neurons: the third generation of neural network models","volume":"10","author":"Maass","year":"1997","journal-title":"Neural Netw."},{"key":"nceacf1c5bib2","doi-asserted-by":"publisher","first-page":"668","DOI":"10.1126\/science.1254642","article-title":"A million spiking-neuron integrated circuit with a scalable communication network and interface","volume":"345","author":"Merolla","year":"2014","journal-title":"Science"},{"key":"nceacf1c5bib3","doi-asserted-by":"publisher","first-page":"82","DOI":"10.1109\/MM.2018.112130359","article-title":"Loihi: a neuromorphic manycore processor with on-chip learning","volume":"38","author":"Davies","year":"2018","journal-title":"IEEE Micro"},{"key":"nceacf1c5bib4","doi-asserted-by":"publisher","first-page":"533","DOI":"10.1038\/323533a0","article-title":"Learning representations by back-propagating errors","volume":"323","author":"Rumelhart","year":"1986","journal-title":"Nature"},{"key":"nceacf1c5bib5","doi-asserted-by":"publisher","first-page":"17","DOI":"10.1016\/S0925-2312(01)00658-0","article-title":"Error-backpropagation in temporally encoded networks of spiking neurons","volume":"48","author":"Bohte","year":"2002","journal-title":"Neurocomputing"},{"key":"nceacf1c5bib6","doi-asserted-by":"publisher","first-page":"552","DOI":"10.1016\/j.ipl.2005.05.023","article-title":"A gradient descent rule for spiking neurons emitting multiple spikes","volume":"95","author":"Booij","year":"2005","journal-title":"Inf. Process. Lett."},{"key":"nceacf1c5bib7","doi-asserted-by":"publisher","first-page":"1419","DOI":"10.1016\/j.neunet.2009.04.003","article-title":"A new supervised learning algorithm for multiple spiking neural networks with application in epilepsy and seizure detection","volume":"22","author":"Ghosh-Dastidar","year":"2009","journal-title":"Neural Netw."},{"key":"nceacf1c5bib8","doi-asserted-by":"publisher","DOI":"10.1142\/S0129065720500276","article-title":"Temporal backpropagation for spiking neural networks with one spike per neuron","volume":"30","author":"Kheradpisheh","year":"2020","journal-title":"Int. J. Neural Syst."},{"key":"nceacf1c5bib9","first-page":"pp 1","article-title":"Fast-classifying, high-accuracy spiking deep networks through weight and threshold balancing","author":"Diehl","year":"2015"},{"key":"nceacf1c5bib10","doi-asserted-by":"publisher","first-page":"178","DOI":"10.3389\/fnins.2013.00178","article-title":"Real-time classification and sensor fusion with a spiking deep belief network","volume":"7","author":"O\u2019Connor","year":"2013","journal-title":"Front. Neurosci."},{"article-title":"Spiking deep networks with LIF neurons","year":"2015","author":"Hunsberger","key":"nceacf1c5bib11"},{"article-title":"Gradient descent for spiking neural networks","year":"2018","author":"Huh","key":"nceacf1c5bib12"},{"article-title":"Slayer: spike layer error reassignment in time","year":"2018","author":"Shrestha","key":"nceacf1c5bib13"},{"article-title":"Hybrid macro\/micro level backpropagation for training deep spiking neural networks","year":"2018","author":"Jin","key":"nceacf1c5bib14"},{"key":"nceacf1c5bib15","doi-asserted-by":"publisher","first-page":"508","DOI":"10.3389\/fnins.2016.00508","article-title":"Training deep spiking neural networks using backpropagation","volume":"10","author":"Lee","year":"2016","journal-title":"Front. Neurosci."},{"key":"nceacf1c5bib16","doi-asserted-by":"publisher","first-page":"119","DOI":"10.3389\/fnins.2020.00119","article-title":"Enabling spike-based backpropagation for training deep neural network architectures","volume":"14","author":"Lee","year":"2020","journal-title":"Front. Neurosci."},{"article-title":"Long short-term memory and learning-to-learn in networks of spiking neurons","year":"2018","author":"Bellec","key":"nceacf1c5bib17"},{"key":"nceacf1c5bib18","article-title":"Spike-train level backpropagation for training deep recurrent spiking neural networks","volume":"32","author":"Zhang","year":"2019"},{"key":"nceacf1c5bib19","doi-asserted-by":"publisher","first-page":"3625","DOI":"10.1038\/s41467-020-17236-y","article-title":"A solution to the learning dilemma for recurrent networks of spiking neurons","volume":"11","author":"Bellec","year":"2020","journal-title":"Nat. Commun."},{"key":"nceacf1c5bib20","doi-asserted-by":"publisher","first-page":"325","DOI":"10.1038\/s42256-020-0187-0","article-title":"Deep learning incorporating biologically inspired neural dynamics and in-memory computing","volume":"2","author":"Wo\u017aniak","year":"2020","journal-title":"Nat. Mach. Intell."},{"key":"nceacf1c5bib21","doi-asserted-by":"publisher","first-page":"39","DOI":"10.1016\/j.neucom.2018.11.014","article-title":"BP-STDP: approximating backpropagation using spike timing dependent plasticity","volume":"330","author":"Tavanaei","year":"2019","journal-title":"Neurocomputing"},{"key":"nceacf1c5bib22","first-page":"pp 367","article-title":"In-hardware learning of multilayer spiking neural networks on a neuromorphic processor","author":"Shrestha","year":"2021"},{"key":"nceacf1c5bib23","first-page":"pp 1","article-title":"Software-level accuracy using stochastic computing with charge-trap-flash based weight matrix","author":"Bhatt","year":"2020"},{"key":"nceacf1c5bib24","first-page":"pp 807","article-title":"Rectified linear units improve restricted Boltzmann machines","author":"Nair","year":"2010"},{"article-title":"SpikeGrad: an ANN-equivalent computation model for implementing backpropagation with spikes","year":"2019","author":"Thiele","key":"nceacf1c5bib25"},{"key":"nceacf1c5bib26","first-page":"pp 1054","article-title":"Hebbian learning with winner take all for spiking neural networks","author":"Gupta","year":"2009"},{"key":"nceacf1c5bib27","first-page":"pp 1756","article-title":"Cross-entropy vs. squared error training: a theoretical and experimental comparison","volume":"vol 13","author":"Golik","year":"2013"},{"key":"nceacf1c5bib28","first-page":"pp 1026","article-title":"Delving deep into rectifiers: surpassing human-level performance on ImageNet classification","author":"He","year":"2015"},{"key":"nceacf1c5bib29","first-page":"pp 2814","article-title":"Understanding dropout","volume":"vol 26","author":"Baldi","year":"2013","journal-title":"Advances in Neural Information Processing Systems"},{"key":"nceacf1c5bib30","doi-asserted-by":"publisher","first-page":"1097","DOI":"10.1145\/3065386","article-title":"ImageNet classification with deep convolutional neural networks","volume":"vol 25","author":"Krizhevsky","year":"2012","journal-title":"Advances in Neural Information Processing Systems"},{"key":"nceacf1c5bib31","doi-asserted-by":"publisher","first-page":"333","DOI":"10.3389\/fnins.2016.00333","article-title":"Acceleration of deep neural network training with resistive cross-point devices: design considerations","volume":"10","author":"Gokmen","year":"2016","journal-title":"Front. Neurosci."},{"key":"nceacf1c5bib32","doi-asserted-by":"publisher","first-page":"2278","DOI":"10.1109\/5.726791","article-title":"Gradient-based learning applied to document recognition","volume":"86","author":"LeCun","year":"1998","journal-title":"Proc. IEEE"},{"article-title":"Fashion-MNIST: a novel image dataset for benchmarking machine learning algorithms","year":"2017","author":"Xiao","key":"nceacf1c5bib33"},{"key":"nceacf1c5bib34","first-page":"pp 2921","article-title":"EMNIST: extending MNIST to handwritten letters","author":"Cohen","year":"2017"},{"key":"nceacf1c5bib35","doi-asserted-by":"publisher","first-page":"437","DOI":"10.3389\/fnins.2015.00437","article-title":"Converting static image datasets to spiking neuromorphic datasets using saccades","volume":"9","author":"Orchard","year":"2015","journal-title":"Front. Neurosci."},{"article-title":"Tonic: event-based datasets and transformations","year":"2021","author":"Lenz","key":"nceacf1c5bib36"},{"key":"nceacf1c5bib37","doi-asserted-by":"publisher","first-page":"45","DOI":"10.1109\/TCSI.2020.3035575","article-title":"Ultra-low-power FDSOI neural circuits for extreme-edge neuromorphic intelligence","volume":"68","author":"Rubino","year":"2020","journal-title":"IEEE Trans. Circuits Syst. I"},{"key":"nceacf1c5bib38","first-page":"pp 275","article-title":"Liquid state machines: motivation, theory and applications","author":"Maass","year":"2011"},{"key":"nceacf1c5bib39","doi-asserted-by":"publisher","first-page":"593","DOI":"10.1016\/j.jcss.2004.04.001","article-title":"On the computational power of circuits of spiking neurons","volume":"69","author":"Maass","year":"2004","journal-title":"J. Comput. Syst. Sci."},{"key":"nceacf1c5bib40","doi-asserted-by":"publisher","first-page":"3212","DOI":"10.1109\/TCSI.2022.3172176","article-title":"Quantum tunneling based ultra-compact and energy efficient spiking neuron enables hardware SNN","volume":"69","author":"Singh","year":"2022","journal-title":"IEEE Trans. Circuits Syst. I"},{"key":"nceacf1c5bib41","first-page":"pp 25703","article-title":"Increasing liquid state machine performance with edge-of-chaos dynamics organized by astrocyte-modulated plasticity","volume":"vol 34","author":"Ivanov","year":"2021","journal-title":"Advances in Neural Information Processing Systems"},{"key":"nceacf1c5bib42","doi-asserted-by":"publisher","first-page":"424","DOI":"10.3389\/fnins.2020.00424","article-title":"Synaptic plasticity dynamics for deep continuous local learning (DECOLLE)","volume":"14","author":"Kaiser","year":"2020","journal-title":"Front. Neurosci."},{"key":"nceacf1c5bib43","doi-asserted-by":"publisher","first-page":"699","DOI":"10.1109\/JPROC.2014.2313565","article-title":"Neurogrid: a mixed-analog-digital multichip system for large-scale neural simulations","volume":"102","author":"Benjamin","year":"2014","journal-title":"Proc. IEEE"},{"key":"nceacf1c5bib44","doi-asserted-by":"publisher","first-page":"911","DOI":"10.1109\/JPROC.2021.3067593","article-title":"Advancing neuromorphic computing with Loihi: a survey of results and outlook","volume":"109","author":"Davies","year":"2021","journal-title":"Proc. IEEE"},{"key":"nceacf1c5bib45","doi-asserted-by":"publisher","first-page":"652","DOI":"10.1109\/JPROC.2014.2304638","article-title":"The spinnaker project","volume":"102","author":"Furber","year":"2014","journal-title":"Proc. IEEE"},{"key":"nceacf1c5bib46","doi-asserted-by":"publisher","DOI":"10.1088\/1741-2560\/13\/5\/051001","article-title":"Large-scale neuromorphic computing systems","volume":"13","author":"Furber","year":"2016","journal-title":"J. Neural Eng."},{"key":"nceacf1c5bib47","first-page":"pp 36","article-title":"Unassisted true analog neural network training chip","author":"Kohda","year":"2020"},{"key":"nceacf1c5bib48","first-page":"pp 422","article-title":"Analog CMOS-based resistive processing unit for deep neural network training","author":"Kim","year":"2017"},{"key":"nceacf1c5bib49","first-page":"pp 1","article-title":"Reliability challenges with materials for analog computing","author":"Cartier","year":"2019"},{"key":"nceacf1c5bib50","first-page":"pp 678","article-title":"A VLSI network of spiking neurons with plastic fully configurable \u2018stop-learning\u2019 synapses","author":"Giulioni","year":"2008"},{"key":"nceacf1c5bib51","first-page":"pp 1044","article-title":"AER-based robotic closed-loop control system","author":"Jim\u00e9nez-Fernandez","year":"2008"},{"key":"nceacf1c5bib52","doi-asserted-by":"publisher","first-page":"693","DOI":"10.1038\/nnano.2016.70","article-title":"Stochastic phase-change neurons","volume":"11","author":"Tuma","year":"2016","journal-title":"Nat. Nanotechnol."},{"key":"nceacf1c5bib53","doi-asserted-by":"publisher","DOI":"10.1063\/1.5108694","article-title":"Pr x Ca 1\u2212x MnO3 based stochastic neuron for Boltzmann machine to solve \u201cmaximum cut\u201d problem","volume":"7","author":"Khilwani","year":"2019","journal-title":"APL Mater."},{"key":"nceacf1c5bib54","doi-asserted-by":"publisher","first-page":"1","DOI":"10.1038\/srep44370","article-title":"Intrinsic optimization using stochastic nanomagnets","volume":"7","author":"Sutton","year":"2017","journal-title":"Sci. Rep."}],"container-title":["Neuromorphic Computing and Engineering"],"original-title":[],"link":[{"URL":"https:\/\/iopscience.iop.org\/article\/10.1088\/2634-4386\/acf1c5","content-type":"text\/html","content-version":"am","intended-application":"text-mining"},{"URL":"https:\/\/iopscience.iop.org\/article\/10.1088\/2634-4386\/acf1c5\/pdf","content-type":"application\/pdf","content-version":"am","intended-application":"text-mining"},{"URL":"https:\/\/iopscience.iop.org\/article\/10.1088\/2634-4386\/acf1c5","content-type":"text\/html","content-version":"vor","intended-application":"text-mining"},{"URL":"https:\/\/iopscience.iop.org\/article\/10.1088\/2634-4386\/acf1c5\/pdf","content-type":"application\/pdf","content-version":"vor","intended-application":"text-mining"},{"URL":"https:\/\/iopscience.iop.org\/article\/10.1088\/2634-4386\/acf1c5\/pdf","content-type":"application\/pdf","content-version":"am","intended-application":"syndication"},{"URL":"https:\/\/iopscience.iop.org\/article\/10.1088\/2634-4386\/acf1c5\/pdf","content-type":"application\/pdf","content-version":"vor","intended-application":"syndication"},{"URL":"https:\/\/iopscience.iop.org\/article\/10.1088\/2634-4386\/acf1c5\/pdf","content-type":"application\/pdf","content-version":"am","intended-application":"similarity-checking"},{"URL":"https:\/\/iopscience.iop.org\/article\/10.1088\/2634-4386\/acf1c5\/pdf","content-type":"application\/pdf","content-version":"vor","intended-application":"similarity-checking"}],"deposited":{"date-parts":[[2023,8,31]],"date-time":"2023-08-31T11:26:47Z","timestamp":1693481207000},"score":1,"resource":{"primary":{"URL":"https:\/\/iopscience.iop.org\/article\/10.1088\/2634-4386\/acf1c5"}},"subtitle":[],"short-title":[],"issued":{"date-parts":[[2023,8,31]]},"references-count":54,"journal-issue":{"issue":"3","published-online":{"date-parts":[[2023,8,31]]},"published-print":{"date-parts":[[2023,9,1]]}},"URL":"https:\/\/doi.org\/10.1088\/2634-4386\/acf1c5","relation":{},"ISSN":["2634-4386"],"issn-type":[{"type":"electronic","value":"2634-4386"}],"subject":[],"published":{"date-parts":[[2023,8,31]]},"assertion":[{"value":"A temporally and spatially local spike-based backpropagation algorithm to enable training in hardware","name":"article_title","label":"Article Title"},{"value":"Neuromorphic Computing and Engineering","name":"journal_title","label":"Journal Title"},{"value":"paper","name":"article_type","label":"Article Type"},{"value":"\u00a9 2023 The Author(s). Published by IOP Publishing Ltd","name":"copyright_information","label":"Copyright Information"},{"value":"2022-03-31","name":"date_received","label":"Date Received","group":{"name":"publication_dates","label":"Publication dates"}},{"value":"2023-08-18","name":"date_accepted","label":"Date Accepted","group":{"name":"publication_dates","label":"Publication dates"}},{"value":"2023-08-31","name":"date_epub","label":"Online publication date","group":{"name":"publication_dates","label":"Publication dates"}}]}}