{"status":"ok","message-type":"work","message-version":"1.0.0","message":{"indexed":{"date-parts":[[2026,5,6]],"date-time":"2026-05-06T15:46:22Z","timestamp":1778082382553,"version":"3.51.4"},"reference-count":67,"publisher":"Proceedings of the National Academy of Sciences","issue":"4","license":[{"start":{"date-parts":[[2022,1,18]],"date-time":"2022-01-18T00:00:00Z","timestamp":1642464000000},"content-version":"vor","delay-in-days":4,"URL":"https:\/\/creativecommons.org\/licenses\/by\/4.0\/"}],"funder":[{"DOI":"10.13039\/501100007601","name":"EC | Horizon 2020","doi-asserted-by":"publisher","award":["720270"],"award-info":[{"award-number":["720270"]}],"id":[{"id":"10.13039\/501100007601","id-type":"DOI","asserted-by":"publisher"}]},{"DOI":"10.13039\/501100007601","name":"EC | Horizon 2020","doi-asserted-by":"publisher","award":["785907"],"award-info":[{"award-number":["785907"]}],"id":[{"id":"10.13039\/501100007601","id-type":"DOI","asserted-by":"publisher"}]},{"DOI":"10.13039\/501100007601","name":"EC | Horizon 2020","doi-asserted-by":"publisher","award":["945539"],"award-info":[{"award-number":["945539"]}],"id":[{"id":"10.13039\/501100007601","id-type":"DOI","asserted-by":"publisher"}]}],"content-domain":{"domain":["www.pnas.org"],"crossmark-restriction":true},"short-container-title":["Proc. Natl. Acad. Sci. U.S.A."],"published-print":{"date-parts":[[2022,1,25]]},"abstract":"<jats:title>Significance<\/jats:title><jats:p>Neuromorphic systems aim to accomplish efficient computation in electronics by mirroring neurobiological principles. Taking advantage of neuromorphic technologies requires effective learning algorithms capable of instantiating high-performing neural networks, while also dealing with inevitable manufacturing variations of individual components, such as memristors or analog neurons. We present a learning framework resulting in bioinspired spiking neural networks with high performance, low inference latency, and sparse spike-coding schemes, which also self-corrects for device mismatch. We validate our approach on the BrainScaleS-2 analog spiking neuromorphic system, demonstrating state-of-the-art accuracy, low latency, and energy efficiency. Our work sketches a path for building powerful neuromorphic processors that take advantage of emerging analog technologies.<\/jats:p>","DOI":"10.1073\/pnas.2109194119","type":"journal-article","created":{"date-parts":[[2022,1,18]],"date-time":"2022-01-18T22:55:20Z","timestamp":1642546520000},"update-policy":"https:\/\/doi.org\/10.1073\/pnas.cm10313","source":"Crossref","is-referenced-by-count":99,"title":["Surrogate gradients for analog neuromorphic computing"],"prefix":"10.1073","volume":"119","author":[{"ORCID":"https:\/\/orcid.org\/0000-0001-7948-4391","authenticated-orcid":false,"given":"Benjamin","family":"Cramer","sequence":"first","affiliation":[{"name":"Kirchhoff-Institute for Physics, Heidelberg University, 69120 Heidelberg, Germany"}]},{"given":"Sebastian","family":"Billaudelle","sequence":"additional","affiliation":[{"name":"Kirchhoff-Institute for Physics, Heidelberg University, 69120 Heidelberg, Germany"}]},{"given":"Simeon","family":"Kanya","sequence":"additional","affiliation":[{"name":"Kirchhoff-Institute for Physics, Heidelberg University, 69120 Heidelberg, Germany"}]},{"given":"Aron","family":"Leibfried","sequence":"additional","affiliation":[{"name":"Kirchhoff-Institute for Physics, Heidelberg University, 69120 Heidelberg, Germany"}]},{"ORCID":"https:\/\/orcid.org\/0000-0002-3955-4815","authenticated-orcid":false,"given":"Andreas","family":"Gr\u00fcbl","sequence":"additional","affiliation":[{"name":"Kirchhoff-Institute for Physics, Heidelberg University, 69120 Heidelberg, Germany"}]},{"given":"Vitali","family":"Karasenko","sequence":"additional","affiliation":[{"name":"Kirchhoff-Institute for Physics, Heidelberg University, 69120 Heidelberg, Germany"}]},{"given":"Christian","family":"Pehle","sequence":"additional","affiliation":[{"name":"Kirchhoff-Institute for Physics, Heidelberg University, 69120 Heidelberg, Germany"}]},{"given":"Korbinian","family":"Schreiber","sequence":"additional","affiliation":[{"name":"Kirchhoff-Institute for Physics, Heidelberg University, 69120 Heidelberg, Germany"}]},{"ORCID":"https:\/\/orcid.org\/0000-0001-9603-6777","authenticated-orcid":false,"given":"Yannik","family":"Stradmann","sequence":"additional","affiliation":[{"name":"Kirchhoff-Institute for Physics, Heidelberg University, 69120 Heidelberg, Germany"}]},{"ORCID":"https:\/\/orcid.org\/0000-0002-1435-0250","authenticated-orcid":false,"given":"Johannes","family":"Weis","sequence":"additional","affiliation":[{"name":"Kirchhoff-Institute for Physics, Heidelberg University, 69120 Heidelberg, Germany"}]},{"given":"Johannes","family":"Schemmel","sequence":"additional","affiliation":[{"name":"Kirchhoff-Institute for Physics, Heidelberg University, 69120 Heidelberg, Germany"}]},{"ORCID":"https:\/\/orcid.org\/0000-0003-1883-644X","authenticated-orcid":false,"given":"Friedemann","family":"Zenke","sequence":"additional","affiliation":[{"name":"Computational Neuroscience Group, Friedrich Miescher Institute for Biomedical Research, 4058 Basel, Switzerland"}]}],"member":"341","published-online":{"date-parts":[[2022,1,14]]},"reference":[{"key":"e_1_3_4_1_2","unstructured":"V. Mnih . Playing Atari with deep reinforcement learning. arXiv [Preprint] (2013). https:\/\/arxiv.org\/abs\/1312.5602 23 (Accessed 23 December 2021)."},{"key":"e_1_3_4_2_2","doi-asserted-by":"publisher","DOI":"10.1038\/nature24270"},{"key":"e_1_3_4_3_2","unstructured":"T. B. Brown . Language models are few-shot learners. arXiv [Preprint] (2020). https:\/\/arxiv.org\/abs\/2005.14165 23 (Accessed 23 December 2021)."},{"key":"e_1_3_4_4_2","doi-asserted-by":"crossref","DOI":"10.7551\/mitpress\/9780262028707.001.0001","volume-title":"Principles of Neural Design","author":"Sterling P.","year":"2015","unstructured":"P. Sterling, S. Laughlin, Principles of Neural Design (MIT Press, Cambridge, MA, 2015)."},{"key":"e_1_3_4_5_2","doi-asserted-by":"publisher","DOI":"10.1109\/5.58356"},{"key":"e_1_3_4_6_2","volume-title":"Analog VLSI Implementation of Neural Systems","author":"Mead C.","year":"2012","unstructured":"C. Mead, M. Ismail, Analog VLSI Implementation of Neural Systems (Springer Science & Business Media, 2012)."},{"key":"e_1_3_4_7_2","doi-asserted-by":"publisher","DOI":"10.3389\/fnins.2011.00073"},{"key":"e_1_3_4_8_2","doi-asserted-by":"publisher","DOI":"10.3389\/fnins.2018.00891"},{"key":"e_1_3_4_9_2","unstructured":"C. D. Schuman . A survey of neuromorphic computing and neural networks in hardware. arXiv [Preprint] (2017). https:\/\/arxiv.org\/abs\/1705.06963 23 (Accessed 23 December 2021)."},{"key":"e_1_3_4_10_2","doi-asserted-by":"publisher","DOI":"10.1126\/science.1254642"},{"key":"e_1_3_4_11_2","doi-asserted-by":"publisher","DOI":"10.1109\/JPROC.2014.2313565"},{"key":"e_1_3_4_12_2","doi-asserted-by":"publisher","DOI":"10.1088\/1741-2560\/13\/5\/051001"},{"key":"e_1_3_4_13_2","doi-asserted-by":"publisher","DOI":"10.1109\/MCSE.2017.33"},{"key":"e_1_3_4_14_2","doi-asserted-by":"publisher","DOI":"10.1109\/MM.2018.112130359"},{"key":"e_1_3_4_15_2","doi-asserted-by":"publisher","DOI":"10.1038\/s41586-019-1677-2"},{"key":"e_1_3_4_16_2","doi-asserted-by":"publisher","DOI":"10.1038\/354515a0"},{"key":"e_1_3_4_17_2","doi-asserted-by":"publisher","DOI":"10.1109\/ISCAS.2010.5536970"},{"key":"e_1_3_4_18_2","doi-asserted-by":"publisher","DOI":"10.1109\/JPROC.2014.2313954"},{"key":"e_1_3_4_19_2","doi-asserted-by":"publisher","DOI":"10.1038\/s42254-020-0208-2"},{"key":"e_1_3_4_20_2","doi-asserted-by":"publisher","DOI":"10.1038\/s41586-018-0180-5"},{"key":"e_1_3_4_21_2","unstructured":"Y. Bengio N. L\u00e9onard A. Courville Estimating or propagating gradients through stochastic neurons for conditional computation. arXiv [Preprint] (2013). https:\/\/arxiv.org\/abs\/1308.3432 (Accessed 23 December 2021)."},{"key":"e_1_3_4_22_2","unstructured":"M. Courbariaux I. Hubara D. Soudry R. El-Yaniv Y. Bengio Binarized neural networks: Training deep neural networks with weights and activations constrained to+ 1 or-1. arXiv [Preprint] (2016). https:\/\/arxiv.org\/abs\/1602.02830 (Accessed 23 December 2021)."},{"key":"e_1_3_4_23_2","doi-asserted-by":"publisher","DOI":"10.1109\/MSP.2019.2931595"},{"key":"e_1_3_4_24_2","doi-asserted-by":"publisher","DOI":"10.3389\/fnins.2021.651141"},{"key":"e_1_3_4_25_2","doi-asserted-by":"publisher","DOI":"10.1038\/s42256-021-00397-w"},{"key":"e_1_3_4_26_2","doi-asserted-by":"publisher","DOI":"10.1109\/ISCAS45731.2020.9180741"},{"key":"e_1_3_4_27_2","doi-asserted-by":"publisher","DOI":"10.1007\/s11265-020-01558-7"},{"key":"e_1_3_4_28_2","unstructured":"J. Schemmel S. Billaudelle P. Dauer J. Weis Accelerated analog neuromorphic computing. arXiv [Preprint] (2020). https:\/\/arxiv.org\/abs\/2003.11996 (Accessed 23 December 2021)."},{"key":"e_1_3_4_29_2","first-page":"8024","volume-title":"Advances in Neural Information Processing Systems","author":"Paszke A.","year":"2019","unstructured":"A. Paszke ., \u201cPytorch: An imperative style, high-performance deep learning library\u201d in Advances in Neural Information Processing Systems, H. Wallach ., Eds. (Curran Associates, Inc., Red Hook, NY, 2019), vol. 32, pp. 8024\u20138035."},{"key":"e_1_3_4_30_2","doi-asserted-by":"publisher","DOI":"10.1109\/TBCAS.2016.2579164"},{"key":"e_1_3_4_31_2","doi-asserted-by":"publisher","DOI":"10.1162\/neco_a_01086"},{"key":"e_1_3_4_32_2","doi-asserted-by":"publisher","DOI":"10.1109\/5.726791"},{"key":"e_1_3_4_33_2","unstructured":"D. P. Kingma J. Ba Adam: A method for stochastic optimization. arXiv [Preprint] (2014). https:\/\/arxiv.org\/abs\/1412.6980 (Accessed 23 December 2021)."},{"key":"e_1_3_4_34_2","unstructured":"H. Xiao K. Rasul R. Vollgraf Fashion-MNIST: A novel image dataset for benchmarking machine learning algorithms. arXiv [Preprint] (2017). https:\/\/arxiv.org\/abs\/1708.07747 (Accessed 23 December 2021)."},{"key":"e_1_3_4_35_2","doi-asserted-by":"publisher","DOI":"10.1162\/neco_a_01367"},{"key":"e_1_3_4_36_2","doi-asserted-by":"publisher","DOI":"10.1109\/TNNLS.2020.3044364"},{"key":"e_1_3_4_37_2","doi-asserted-by":"publisher","DOI":"10.1145\/3320288.3320304"},{"key":"e_1_3_4_38_2","doi-asserted-by":"publisher","DOI":"10.1109\/ISCAS45731.2020.9180440"},{"key":"e_1_3_4_39_2","doi-asserted-by":"publisher","DOI":"10.1073\/pnas.1604850113"},{"key":"e_1_3_4_40_2","doi-asserted-by":"publisher","DOI":"10.3389\/fnins.2018.00774"},{"key":"e_1_3_4_41_2","doi-asserted-by":"publisher","DOI":"10.1016\/S0925-2312(01)00658-0"},{"key":"e_1_3_4_42_2","unstructured":"B. R\u00fcckauer N. K\u00e4nzig S. C. Liu T. Delbruck Y. Sandamirskaya Closing the accuracy gap in an event-based visual recognition task. arXiv Preprint (2019). https:\/\/arxiv.org\/abs\/1906.08859 (Accessed 23 December 2021)."},{"key":"e_1_3_4_43_2","doi-asserted-by":"publisher","DOI":"10.3389\/fnins.2018.00987"},{"key":"e_1_3_4_44_2","doi-asserted-by":"publisher","DOI":"10.1038\/s41598-021-02779-x"},{"key":"e_1_3_4_45_2","unstructured":"E. Hunsberger C. Eliasmith Spiking deep networks with LIF neurons. arXiv [Preprint] (2015). https:\/\/arxiv.org\/abs\/1510.08829 (Accessed 23 December 2021)."},{"key":"e_1_3_4_46_2","doi-asserted-by":"publisher","DOI":"10.3389\/fnins.2016.00508"},{"key":"e_1_3_4_47_2","first-page":"3227","article-title":"Supervised learning based on temporal coding in spiking neural networks","volume":"29","author":"Mostafa H.","year":"2018","unstructured":"H. Mostafa, Supervised learning based on temporal coding in spiking neural networks. IEEE Trans. Neural Netw. Learn. Syst. 29, 3227\u20133235 (2018).","journal-title":"IEEE Trans. Neural Netw. Learn. Syst."},{"key":"e_1_3_4_48_2","doi-asserted-by":"publisher","DOI":"10.1038\/s41467-020-17236-y"},{"key":"e_1_3_4_49_2","first-page":"1","volume-title":"Advances in Neural Information Processing Systems","author":"Huh D.","year":"2018","unstructured":"D. Huh, T. J. Sejnowski, \u201cGradient descent for spiking neural networks\u201d in Advances in Neural Information Processing Systems, S. Bengio ., Eds. (Curran Associates, Inc., Red Hook, NY, 2018), vol. 31, pp. 1\u201311."},{"key":"e_1_3_4_50_2","doi-asserted-by":"publisher","DOI":"10.1038\/s41467-020-16108-9"},{"key":"e_1_3_4_51_2","doi-asserted-by":"publisher","DOI":"10.1038\/s41928-020-00523-3"},{"key":"e_1_3_4_52_2","doi-asserted-by":"publisher","DOI":"10.1109\/TNNLS.2012.2204770"},{"key":"e_1_3_4_53_2","doi-asserted-by":"publisher","DOI":"10.1109\/SOCC.2012.6398336"},{"key":"e_1_3_4_54_2","doi-asserted-by":"publisher","DOI":"10.1109\/JETCAS.2016.2547718"},{"key":"e_1_3_4_55_2","doi-asserted-by":"publisher","DOI":"10.1109\/TVLSI.2019.2893256"},{"key":"e_1_3_4_56_2","doi-asserted-by":"publisher","DOI":"10.1109\/IJCNN.2017.7966125"},{"key":"e_1_3_4_57_2","doi-asserted-by":"publisher","DOI":"10.1038\/s42256-021-00388-x"},{"key":"e_1_3_4_58_2","doi-asserted-by":"publisher","DOI":"10.1038\/s41586-020-1942-4"},{"key":"e_1_3_4_59_2","unstructured":"L. G. Wright . Deep physical neural networks enabled by a backpropagation algorithm for arbitrary physical systems. arXiv [Preprint] (2021). https:\/\/arxiv.org\/abs\/2104.13386 (Accessed 23 December 2021)."},{"key":"e_1_3_4_60_2","doi-asserted-by":"publisher","DOI":"10.1038\/s42256-019-0097-1"},{"key":"e_1_3_4_61_2","doi-asserted-by":"publisher","DOI":"10.3389\/fnins.2015.00222"},{"key":"e_1_3_4_62_2","first-page":"1117","article-title":"Backpropagation for energy-efficient neuromorphic computing","volume":"28","author":"Esser S. K.","year":"2015","unstructured":"S. K. Esser, R. Appuswamy, P. Merolla, J. V. Arthur, D. S. Modha, Backpropagation for energy-efficient neuromorphic computing. Adv. Neural Inf. Process. Syst. 28, 1117\u20131125 (2015).","journal-title":"Adv. Neural Inf. Process. Syst."},{"key":"e_1_3_4_63_2","doi-asserted-by":"publisher","DOI":"10.1109\/JSSC.2018.2884901"},{"key":"e_1_3_4_64_2","doi-asserted-by":"publisher","DOI":"10.1109\/TBCAS.2019.2928793"},{"key":"e_1_3_4_65_2","doi-asserted-by":"publisher","DOI":"10.1038\/s41467-021-26022-3"},{"key":"e_1_3_4_66_2","unstructured":"E. M\u00fcller . Extending brainscales OS for brainscales-2. arXiv [Preprint] (2020). https:\/\/arxiv.org\/abs\/2003.13750 (Accessed 23 December 2021)."},{"key":"e_1_3_4_67_2","doi-asserted-by":"publisher","DOI":"10.1109\/ICCV.2015.123"}],"container-title":["Proceedings of the National Academy of Sciences"],"original-title":[],"language":"en","link":[{"URL":"http:\/\/www.pnas.org\/syndication\/doi\/10.1073\/pnas.2109194119","content-type":"unspecified","content-version":"vor","intended-application":"syndication"},{"URL":"https:\/\/pnas.org\/doi\/pdf\/10.1073\/pnas.2109194119","content-type":"unspecified","content-version":"vor","intended-application":"similarity-checking"}],"deposited":{"date-parts":[[2024,9,16]],"date-time":"2024-09-16T15:10:35Z","timestamp":1726499435000},"score":1,"resource":{"primary":{"URL":"https:\/\/pnas.org\/doi\/full\/10.1073\/pnas.2109194119"}},"subtitle":[],"short-title":[],"issued":{"date-parts":[[2022,1,14]]},"references-count":67,"journal-issue":{"issue":"4","published-print":{"date-parts":[[2022,1,25]]}},"alternative-id":["10.1073\/pnas.2109194119"],"URL":"https:\/\/doi.org\/10.1073\/pnas.2109194119","relation":{},"ISSN":["0027-8424","1091-6490"],"issn-type":[{"value":"0027-8424","type":"print"},{"value":"1091-6490","type":"electronic"}],"subject":[],"published":{"date-parts":[[2022,1,14]]},"assertion":[{"value":"2021-05-19","order":0,"name":"received","label":"Received","group":{"name":"publication_history","label":"Publication History"}},{"value":"2021-11-25","order":1,"name":"accepted","label":"Accepted","group":{"name":"publication_history","label":"Publication History"}},{"value":"2022-01-14","order":2,"name":"published","label":"Published","group":{"name":"publication_history","label":"Publication History"}}],"article-number":"e2109194119"}}