{"status":"ok","message-type":"work","message-version":"1.0.0","message":{"indexed":{"date-parts":[[2026,6,27]],"date-time":"2026-06-27T03:14:24Z","timestamp":1782530064742,"version":"3.54.5"},"reference-count":37,"publisher":"Association for Computing Machinery (ACM)","issue":"6","license":[{"start":{"date-parts":[[2019,11,15]],"date-time":"2019-11-15T00:00:00Z","timestamp":1573776000000},"content-version":"vor","delay-in-days":0,"URL":"https:\/\/www.acm.org\/publications\/policies\/copyright_policy#Background"}],"funder":[{"name":"C-BRIC, one of six centers in JUMP, a Semiconductor Research Corporation"},{"DOI":"10.13039\/100000185","name":"DARPA","doi-asserted-by":"crossref","id":[{"id":"10.13039\/100000185","id-type":"DOI","asserted-by":"crossref"}]}],"content-domain":{"domain":["dl.acm.org"],"crossmark-restriction":true},"short-container-title":["ACM Trans. Embed. Comput. Syst."],"published-print":{"date-parts":[[2019,11,30]]},"abstract":"<jats:p>Resistive crossbars have shown strong potential as the building blocks of future neural fabrics, due to their ability to natively execute vector-matrix multiplication (the dominant computational kernel in DNNs). However, a key challenge that arises in resistive crossbars is that non-idealities in the synaptic devices, interconnects, and peripheral circuits of resistive crossbars lead to errors in the computations performed. When large-scale DNNs are executed on resistive crossbar systems, these errors compound and result in unacceptable degradation in application-level accuracy. We propose CxDNN, a hardware-software methodology that enables the realization of large-scale DNNs on crossbar systems by compensating for errors due to non-idealities, greatly mitigating the degradation in accuracy. CxDNN is composed of (i) an optimized mapping technique to convert floating-point weights and activations to crossbar conductances and input voltages, (ii) a fast one-time re-training method to recover accuracy loss due to this conversion, and (iii) low-overhead compensation hardware to mitigate dynamic and hardware-instance-specific errors. Unlike previous efforts that are limited to small networks and require the training and deployment of hardware-instance-specific models, CxDNN presents a scalable compensation methodology that can address large DNNs (e.g., ResNet-50 on ImageNet) and maintains the train-once-deploy-anywhere tenet of current DNN application. We evaluated CxDNN on six top DNNs on the ImageNet dataset with 0.5--13.8 million neurons and 0.5--15.5 billion connections. CxDNN achieves 16.9%--49% improvement in the top-1 classification accuracy, effectively mitigating a key challenge to the use of resistive crossbar--based neural fabrics.<\/jats:p>","DOI":"10.1145\/3362035","type":"journal-article","created":{"date-parts":[[2019,11,15]],"date-time":"2019-11-15T21:16:57Z","timestamp":1573852617000},"page":"1-23","update-policy":"https:\/\/doi.org\/10.1145\/crossmark-policy","source":"Crossref","is-referenced-by-count":58,"title":["CxDNN"],"prefix":"10.1145","volume":"18","author":[{"ORCID":"https:\/\/orcid.org\/0000-0002-2291-7712","authenticated-orcid":false,"given":"Shubham","family":"Jain","sequence":"first","affiliation":[{"name":"School of Electrical and Computer Engineering, Purdue University, West Lafayette, IN"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Anand","family":"Raghunathan","sequence":"additional","affiliation":[{"name":"School of Electrical and Computer Engineering, Purdue University, West Lafayette, IN"}],"role":[{"vocabulary":"crossref","role":"author"}]}],"member":"320","published-online":{"date-parts":[[2019,11,15]]},"reference":[{"key":"e_1_2_1_1_1","doi-asserted-by":"publisher","DOI":"10.1109\/T-C.1971.223134"},{"key":"e_1_2_1_2_1","first-page":"5","article-title":"Technology aware training in memristive neuromorphic systems for nonideal synaptic crossbars","volume":"2","author":"Chakraborty I.","year":"2018","journal-title":"IEEE Trans. Emerg. Topics. Comput. Intell."},{"key":"e_1_2_1_3_1","volume-title":"Proceedings of the Design, Automation Test in Europe Conference Exhibition (DATE\u201917)","author":"Chen L.","year":"2017"},{"key":"e_1_2_1_4_1","doi-asserted-by":"crossref","unstructured":"P. Chen X. Peng and S. Yu. 2018. NeuroSim: A circuit-level macro model for benchmarking neuro-inspired architectures in online learning. IEEE Trans. Comput.-Aided Des. Integ. Circ. Syst. (2018) 1--1. DOI:https:\/\/doi.org\/10.1109\/TCAD.2018.2789723  P. Chen X. Peng and S. Yu. 2018. NeuroSim: A circuit-level macro model for benchmarking neuro-inspired architectures in online learning. IEEE Trans. Comput.-Aided Des. Integ. Circ. Syst. (2018) 1--1. DOI:https:\/\/doi.org\/10.1109\/TCAD.2018.2789723","DOI":"10.1109\/TCAD.2018.2789723"},{"key":"e_1_2_1_5_1","doi-asserted-by":"publisher","DOI":"10.5555\/2840819.2840848"},{"key":"e_1_2_1_6_1","doi-asserted-by":"publisher","DOI":"10.1145\/3061639.3062326"},{"key":"e_1_2_1_7_1","volume-title":"Proceedings of the ACM\/IEEE 43rd International Symposium on Computer Architecture (ISCA\u201916)","author":"Chi P.","year":"2016"},{"key":"e_1_2_1_8_1","volume-title":"Proceedings of the IEEE International Symposium on High Performance Computer Architecture (HPCA\u201918)","author":"Feinberg B.","year":"2018"},{"key":"e_1_2_1_9_1","doi-asserted-by":"publisher","DOI":"10.3389\/fnins.2017.00538"},{"key":"e_1_2_1_10_1","doi-asserted-by":"publisher","DOI":"10.3389\/fnins.2016.00333"},{"key":"e_1_2_1_11_1","volume-title":"Dally","author":"Han Song","year":"2015"},{"key":"e_1_2_1_12_1","volume-title":"ADC: Automated deep compression and acceleration with reinforcement learning. CoRR abs\/1802.03494","author":"He Yihui","year":"2018"},{"key":"e_1_2_1_13_1","doi-asserted-by":"publisher","DOI":"10.1002\/adma.201705914"},{"key":"e_1_2_1_14_1","volume-title":"Proceedings of the 53rd ACM\/EDAC\/IEEE Design Automation Conference (DAC\u201916)","author":"Hu M."},{"key":"e_1_2_1_15_1","volume-title":"Rx-Caffe: Framework for evaluating and training deep neural networks on resistive crossbars. CoRR abs\/1809.00072","author":"Jain Shubham","year":"2018"},{"key":"e_1_2_1_16_1","doi-asserted-by":"publisher","DOI":"10.1145\/3195970.3196012"},{"key":"e_1_2_1_17_1","volume-title":"Proceedings of the 56th ACM\/IEEE Design Automation Conference (DAC\u201919)","author":"Jain S."},{"key":"e_1_2_1_18_1","doi-asserted-by":"publisher","DOI":"10.1021\/nl904092h"},{"key":"e_1_2_1_19_1","doi-asserted-by":"publisher","DOI":"10.1109\/IJCNN.2015.7280785"},{"key":"e_1_2_1_20_1","volume-title":"Proceedings of the IEEE International Solid-State Circuits Conference Digest of Technical Papers. 468--469","author":"Kull L.","year":"2013"},{"key":"e_1_2_1_21_1","doi-asserted-by":"publisher","DOI":"10.1145\/2463209.2488741"},{"key":"e_1_2_1_22_1","volume-title":"Proceedings of the IEEE\/ACM International Conference on Computer-Aided Design (ICCAD\u201914)","author":"Liu B.","year":"2014"},{"key":"e_1_2_1_23_1","volume-title":"Proceedings of the 54th ACM\/EDAC\/IEEE Design Automation Conference (DAC\u201917)","author":"Liu C."},{"key":"e_1_2_1_24_1","article-title":"Process and electrical results for the on-die interconnect stack for Intel\u2019s 45nm process generation","volume":"12","author":"Moon Peter","year":"2008","journal-title":"Intel Technol. J."},{"key":"e_1_2_1_25_1","volume-title":"Proceedings of the ACM\/IEEE 45th International Symposium on Computer Architecture (ISCA\u201918)","author":"Park E.","year":"2018"},{"key":"e_1_2_1_26_1","unstructured":"R. Parloff. 2016. Why deep learning is suddenly changing your life. Fortune.com. 9\/28\/16. Retrieved from: http:\/\/fortune.com\/ai-artificial-intelligence-deep-machine-learning\/  R. Parloff. 2016. Why deep learning is suddenly changing your life. Fortune.com. 9\/28\/16. Retrieved from: http:\/\/fortune.com\/ai-artificial-intelligence-deep-machine-learning\/"},{"key":"e_1_2_1_27_1","unstructured":"Design Rules. [n.d.]. MOSIS scalable CMOS (SCMOS). Retrieved from: https:\/\/www.mosis.com\/files\/scmos\/scmos.pdf.  Design Rules. [n.d.]. MOSIS scalable CMOS (SCMOS). Retrieved from: https:\/\/www.mosis.com\/files\/scmos\/scmos.pdf."},{"key":"e_1_2_1_28_1","doi-asserted-by":"publisher","DOI":"10.1109\/TBCAS.2016.2525823"},{"key":"e_1_2_1_29_1","volume-title":"Proceedings of the ACM\/IEEE 43rd International Symposium on Computer Architecture (ISCA\u201916)","author":"Shafiee A.","year":"2016"},{"key":"e_1_2_1_30_1","doi-asserted-by":"publisher","DOI":"10.1145\/3061639.3062259"},{"key":"e_1_2_1_31_1","doi-asserted-by":"publisher","DOI":"10.1109\/TBCAS.2015.2414423"},{"key":"e_1_2_1_32_1","doi-asserted-by":"publisher","DOI":"10.1145\/3061639.3062256"},{"key":"e_1_2_1_33_1","volume-title":"Metal-oxide RRAM. Proc. IEEE 100","author":"Philip Wong H. S","year":"2012"},{"key":"e_1_2_1_34_1","volume-title":"Proceedings of the Design, Automation Test in Europe Conference Exhibition (DATE\u201916)","author":"Xia L."},{"key":"e_1_2_1_35_1","volume-title":"Proceedings of the IEEE\/ACM International Conference on Computer-Aided Design (ICCAD\u201917)","author":"Yan B.","year":"2017"},{"key":"e_1_2_1_36_1","volume-title":"Proceedings of the IEEE Symposium on VLSI Circuits (VLSI-Circuits\u201916)","author":"Zhang Jintao","year":"2016"},{"key":"e_1_2_1_37_1","volume-title":"DoReFa-Net: Training low bitwidth convolutional neural networks with low bitwidth gradients. CoRR abs\/1606.06160","author":"Zhou Shuchang","year":"2016"}],"container-title":["ACM Transactions on Embedded Computing Systems"],"original-title":[],"language":"en","link":[{"URL":"https:\/\/dl.acm.org\/doi\/10.1145\/3362035","content-type":"unspecified","content-version":"vor","intended-application":"text-mining"},{"URL":"https:\/\/dl.acm.org\/doi\/pdf\/10.1145\/3362035","content-type":"application\/pdf","content-version":"vor","intended-application":"syndication"},{"URL":"https:\/\/dl.acm.org\/doi\/pdf\/10.1145\/3362035","content-type":"unspecified","content-version":"vor","intended-application":"similarity-checking"}],"deposited":{"date-parts":[[2025,6,17]],"date-time":"2025-06-17T23:44:53Z","timestamp":1750203893000},"score":1,"resource":{"primary":{"URL":"https:\/\/dl.acm.org\/doi\/10.1145\/3362035"}},"subtitle":["Hardware-software Compensation Methods for Deep Neural Networks on Resistive Crossbar Systems"],"short-title":[],"issued":{"date-parts":[[2019,11,15]]},"references-count":37,"journal-issue":{"issue":"6","published-print":{"date-parts":[[2019,11,30]]}},"alternative-id":["10.1145\/3362035"],"URL":"https:\/\/doi.org\/10.1145\/3362035","relation":{},"ISSN":["1539-9087","1558-3465"],"issn-type":[{"value":"1539-9087","type":"print"},{"value":"1558-3465","type":"electronic"}],"subject":[],"published":{"date-parts":[[2019,11,15]]},"assertion":[{"value":"2019-08-01","order":0,"name":"received","label":"Received","group":{"name":"publication_history","label":"Publication History"}},{"value":"2019-09-01","order":1,"name":"accepted","label":"Accepted","group":{"name":"publication_history","label":"Publication History"}},{"value":"2019-11-15","order":2,"name":"published","label":"Published","group":{"name":"publication_history","label":"Publication History"}}]}}