{"status":"ok","message-type":"work","message-version":"1.0.0","message":{"indexed":{"date-parts":[[2026,1,9]],"date-time":"2026-01-09T22:42:06Z","timestamp":1767998526160,"version":"3.49.0"},"reference-count":52,"publisher":"Association for Computing Machinery (ACM)","issue":"5s","license":[{"start":{"date-parts":[[2021,9,22]],"date-time":"2021-09-22T00:00:00Z","timestamp":1632268800000},"content-version":"vor","delay-in-days":0,"URL":"https:\/\/www.acm.org\/publications\/policies\/copyright_policy#Background"}],"funder":[{"DOI":"10.13039\/100000028","name":"Semiconductor Research Corporation","doi-asserted-by":"crossref","award":["#2964.001"],"award-info":[{"award-number":["#2964.001"]}],"id":[{"id":"10.13039\/100000028","id-type":"DOI","asserted-by":"crossref"}]},{"name":"National Science Foundation","award":["#1909854 and #2011236"],"award-info":[{"award-number":["#1909854 and #2011236"]}]}],"content-domain":{"domain":["dl.acm.org"],"crossmark-restriction":true},"short-container-title":["ACM Trans. Embed. Comput. Syst."],"published-print":{"date-parts":[[2021,10,31]]},"abstract":"<jats:p>Hardware accelerators are essential to the accommodation of ever-increasing Deep Neural Network (DNN) workloads on the resource-constrained embedded devices. While accelerators facilitate fast and energy-efficient DNN operations, their accuracy is threatened by faults in their on-chip and off-chip memories, where millions of DNN weights are held. The use of emerging Non-Volatile Memories (NVM) further exposes DNN accelerators to a non-negligible rate of permanent defects due to immature fabrication, limited endurance, and aging. To tolerate defects in NVM-based DNN accelerators, previous work either requires extra redundancy in hardware or performs defect-aware retraining, imposing significant overhead. In comparison, this paper proposes a set of algorithms that exploit the flexibility in setting the fault-free bits in weight memory to effectively approximate weight values, so as to mitigate defect-induced accuracy drop. These algorithms can be applied as a one-step solution when loading the weights to embedded devices. They only require trivial hardware support and impose negligible run-time overhead. Experiments on popular DNN models show that the proposed techniques successfully boost inference accuracy even in the face of elevated defect rates in the weight memory.<\/jats:p>","DOI":"10.1145\/3477016","type":"journal-article","created":{"date-parts":[[2021,9,22]],"date-time":"2021-09-22T20:48:40Z","timestamp":1632343720000},"page":"1-21","update-policy":"https:\/\/doi.org\/10.1145\/crossmark-policy","source":"Crossref","is-referenced-by-count":9,"title":["Tolerating Defects in Low-Power Neural Network Accelerators Via Retraining-Free Weight Approximation"],"prefix":"10.1145","volume":"20","author":[{"given":"Fateme S.","family":"Hosseini","sequence":"first","affiliation":[{"name":"University of Delaware, Newark, USA"}]},{"given":"Fanruo","family":"Meng","sequence":"additional","affiliation":[{"name":"University of Delaware, Newark, USA"}]},{"given":"Chengmo","family":"Yang","sequence":"additional","affiliation":[{"name":"University of Delaware, Newark, USA"}]},{"given":"Wujie","family":"Wen","sequence":"additional","affiliation":[{"name":"Lehigh University, Bethlehem, USA"}]},{"given":"Rosario","family":"Cammarota","sequence":"additional","affiliation":[{"name":"Intel, San Jose, USA"}]}],"member":"320","published-online":{"date-parts":[[2021,9,22]]},"reference":[{"key":"e_1_2_1_1_1","doi-asserted-by":"publisher","DOI":"10.1557\/adv.2016.377"},{"key":"e_1_2_1_2_1","doi-asserted-by":"publisher","DOI":"10.1109\/TED.2015.2439635"},{"key":"e_1_2_1_3_1","doi-asserted-by":"publisher","DOI":"10.1109\/TC.2014.12"},{"key":"e_1_2_1_4_1","doi-asserted-by":"publisher","DOI":"10.5555\/3130379.3130384"},{"key":"e_1_2_1_5_1","unstructured":"F. Chollet et\u00a0al. 2015. Keras. https:\/\/keras.io. (2015).  F. Chollet et\u00a0al. 2015. Keras. https:\/\/keras.io. (2015)."},{"key":"e_1_2_1_6_1","volume-title":"IEEE Workshop on Silicon Errors in Logic-System Effects (SELSE).","author":"DeBardeleben N.","unstructured":"N. DeBardeleben , S. Blanchard , V. Sridharan , S. Gurumurthi , J. Stearley , K. Ferreira , and J. Shalf . 2014. Extra Bits on SRAM and DRAM Errors\u2013More Data From the field . In IEEE Workshop on Silicon Errors in Logic-System Effects (SELSE). N. DeBardeleben, S. Blanchard, V. Sridharan, S. Gurumurthi, J. Stearley, K. Ferreira, and J. Shalf. 2014. Extra Bits on SRAM and DRAM Errors\u2013More Data From the field. In IEEE Workshop on Silicon Errors in Logic-System Effects (SELSE)."},{"key":"e_1_2_1_7_1","doi-asserted-by":"publisher","DOI":"10.1109\/CVPR.2009.5206848"},{"key":"e_1_2_1_8_1","doi-asserted-by":"publisher","DOI":"10.1145\/1629911.1630086"},{"key":"e_1_2_1_9_1","doi-asserted-by":"publisher","DOI":"10.1109\/MM.2019.2944782"},{"key":"e_1_2_1_10_1","doi-asserted-by":"publisher","DOI":"10.1145\/3093337.3037702"},{"key":"e_1_2_1_11_1","doi-asserted-by":"publisher","DOI":"10.1007\/BF00972518"},{"key":"e_1_2_1_12_1","doi-asserted-by":"publisher","DOI":"10.1109\/LCA.2016.2599513"},{"key":"e_1_2_1_13_1","doi-asserted-by":"publisher","DOI":"10.5555\/3361338.3361373"},{"key":"e_1_2_1_14_1","volume-title":"2017 IEEE\/ACM International Symposium on Low Power Electronics and Design (ISLPED). 1\u20136.","author":"Jiang L.","unstructured":"L. Jiang , M. Kim , W. Wen , and D. Wang . 2017. XNOR-POP: A processing-in-memory architecture for binary convolutional neural networks in wide-IO2 DRAMs . In 2017 IEEE\/ACM International Symposium on Low Power Electronics and Design (ISLPED). 1\u20136. L. Jiang, M. Kim, W. Wen, and D. Wang. 2017. XNOR-POP: A processing-in-memory architecture for binary convolutional neural networks in wide-IO2 DRAMs. In 2017 IEEE\/ACM International Symposium on Low Power Electronics and Design (ISLPED). 1\u20136."},{"key":"e_1_2_1_15_1","doi-asserted-by":"publisher","DOI":"10.1109\/TVLSI.2015.2420954"},{"key":"e_1_2_1_16_1","doi-asserted-by":"publisher","DOI":"10.1145\/3140659.3080246"},{"key":"e_1_2_1_17_1","doi-asserted-by":"publisher","DOI":"10.1109\/TCAD.2015.2394434"},{"key":"e_1_2_1_18_1","doi-asserted-by":"publisher","DOI":"10.1109\/TCAD.2016.2615845"},{"key":"e_1_2_1_19_1","doi-asserted-by":"publisher","DOI":"10.1109\/12.21144"},{"key":"e_1_2_1_20_1","unstructured":"A. Krizhevsky V. Nair and G. Hinton. 2010. CIFAR-10 (canadian institute for advanced research). (2010). http:\/\/www.cs.toronto.edu\/ kriz\/cifar.html.  A. Krizhevsky V. Nair and G. Hinton. 2010. CIFAR-10 (canadian institute for advanced research). (2010). http:\/\/www.cs.toronto.edu\/ kriz\/cifar.html."},{"key":"e_1_2_1_21_1","unstructured":"H. T. Kung and M. S. Lam. 1983. Fault-tolerance and two-level pipelining in VLSI systolic arrays.  H. T. Kung and M. S. Lam. 1983. Fault-tolerance and two-level pipelining in VLSI systolic arrays."},{"key":"e_1_2_1_22_1","unstructured":"Y. LeCun and C. Cortes. 2010. MNIST handwritten digit database. (2010). http:\/\/yann.lecun.com\/exdb\/mnist\/.  Y. LeCun and C. Cortes. 2010. MNIST handwritten digit database. (2010). http:\/\/yann.lecun.com\/exdb\/mnist\/."},{"key":"e_1_2_1_23_1","doi-asserted-by":"publisher","DOI":"10.1109\/TC.2013.98"},{"key":"e_1_2_1_24_1","volume-title":"Proceedings of the 50th Annual IEEE\/ACM International Symposium on Microarchitecture (MICRO). 288\u2013301","author":"Li S.","unstructured":"S. Li , D. Niu , K. T. Malladi , H. Zheng , B. Brennan , and Y. Xie . 2017. DRISA: A DRAM-based reconfigurable in-situ accelerator . In Proceedings of the 50th Annual IEEE\/ACM International Symposium on Microarchitecture (MICRO). 288\u2013301 . S. Li, D. Niu, K. T. Malladi, H. Zheng, B. Brennan, and Y. Xie. 2017. DRISA: A DRAM-based reconfigurable in-situ accelerator. In Proceedings of the 50th Annual IEEE\/ACM International Symposium on Microarchitecture (MICRO). 288\u2013301."},{"key":"e_1_2_1_25_1","doi-asserted-by":"publisher","DOI":"10.1145\/2744769.2744930"},{"key":"e_1_2_1_26_1","doi-asserted-by":"publisher","DOI":"10.1145\/3061639.3062310"},{"key":"e_1_2_1_27_1","volume-title":"Automation & Test in Europe Conference & Exhibition (DATE).","author":"Long Y.","unstructured":"Y. Long , X. She , and S. Mukhopadhyay . 2019. Design of Reliable DNN accelerator with unreliable ReRAM. In 2019 Design , Automation & Test in Europe Conference & Exhibition (DATE). Y. Long, X. She, and S. Mukhopadhyay. 2019. Design of Reliable DNN accelerator with unreliable ReRAM. In 2019 Design, Automation & Test in Europe Conference & Exhibition (DATE)."},{"key":"e_1_2_1_28_1","unstructured":"S. Longofono D. Kline R. G. Melhem and A. K. Jones. 2020. A CASTLE with TOWERs for Reliable Secure PCM. IEEE Trans. Comput. (2020) 1\u20131.  S. Longofono D. Kline R. G. Melhem and A. K. Jones. 2020. A CASTLE with TOWERs for Reliable Secure PCM. IEEE Trans. Comput. (2020) 1\u20131."},{"key":"e_1_2_1_29_1","doi-asserted-by":"publisher","DOI":"10.1109\/TC.2020.3000218"},{"key":"e_1_2_1_30_1","doi-asserted-by":"publisher","DOI":"10.1109\/HPCA.2005.37"},{"key":"e_1_2_1_31_1","doi-asserted-by":"crossref","unstructured":"Y. Pan P. Ouyang Y. Zhao W. Kang S. Yin Y. Zhang W. Zhao and S. Wei. 2018. A Multilevel Cell STT-MRAM-based computing in-memory accelerator for binary convolutional neural network. IEEE Transactions on Magnetics (2018) 1\u20135.  Y. Pan P. Ouyang Y. Zhao W. Kang S. Yin Y. Zhang W. Zhao and S. Wei. 2018. A Multilevel Cell STT-MRAM-based computing in-memory accelerator for binary convolutional neural network. IEEE Transactions on Magnetics (2018) 1\u20135.","DOI":"10.1109\/TMAG.2018.2848625"},{"key":"e_1_2_1_32_1","volume-title":"The 2nd International Conference on Next Generation Information Technology. 82\u201385","author":"Park Y.","unstructured":"Y. Park , D. Shin , S. K. Park , and K. H. Park . 2011. Power-aware memory management for hybrid main memory . In The 2nd International Conference on Next Generation Information Technology. 82\u201385 . Y. Park, D. Shin, S. K. Park, and K. H. Park. 2011. Power-aware memory management for hybrid main memory. In The 2nd International Conference on Next Generation Information Technology. 82\u201385."},{"key":"e_1_2_1_33_1","doi-asserted-by":"publisher","DOI":"10.1145\/3352460.3358258"},{"key":"e_1_2_1_34_1","volume-title":"Proceedings of the IEEE International Conference on Computer Vision. 1211\u20131220","author":"Rakin A. S.","unstructured":"A. S. Rakin , Z. He , and D. Fan . 2019. Bit-flip Attack: Crushing neural network with progressive bit search . In Proceedings of the IEEE International Conference on Computer Vision. 1211\u20131220 . A. S. Rakin, Z. He, and D. Fan. 2019. Bit-flip Attack: Crushing neural network with progressive bit search. In Proceedings of the IEEE International Conference on Computer Vision. 1211\u20131220."},{"key":"e_1_2_1_35_1","doi-asserted-by":"publisher","DOI":"10.1145\/3195970.3195997"},{"key":"e_1_2_1_36_1","volume-title":"FAULTSIM: A Fast, configurable memory-resilience simulator. In The Memory Forum: In Conjunction with ISCA","author":"Roberts D.","year":"2014","unstructured":"D. Roberts and P. Nair . 2014 . FAULTSIM: A Fast, configurable memory-resilience simulator. In The Memory Forum: In Conjunction with ISCA , Vol. 41 . D. Roberts and P. Nair. 2014. FAULTSIM: A Fast, configurable memory-resilience simulator. In The Memory Forum: In Conjunction with ISCA, Vol. 41."},{"key":"e_1_2_1_37_1","doi-asserted-by":"publisher","DOI":"10.5555\/2971808.2972024"},{"key":"e_1_2_1_38_1","doi-asserted-by":"publisher","DOI":"10.1145\/1816038.1815980"},{"key":"e_1_2_1_39_1","volume-title":"1990 IJCNN International Joint Conference on Neural Networks.","author":"Sequin C. H.","unstructured":"C. H. Sequin and R. D. Clay . 1990. Fault tolerance in artificial neural networks . In 1990 IJCNN International Joint Conference on Neural Networks. C. H. Sequin and R. D. Clay. 1990. Fault tolerance in artificial neural networks. In 1990 IJCNN International Joint Conference on Neural Networks."},{"key":"e_1_2_1_40_1","doi-asserted-by":"publisher","DOI":"10.1145\/3007787.3001139"},{"key":"e_1_2_1_41_1","volume-title":"2017 IEEE International Symposium on High Performance Computer Architecture (HPCA). 541\u2013552","author":"Song L.","unstructured":"L. Song , X. Qian , H. Li , and Y. Chen . 2017. PipeLayer: A Pipelined ReRAM-Based accelerator for deep learning . In 2017 IEEE International Symposium on High Performance Computer Architecture (HPCA). 541\u2013552 . L. Song, X. Qian, H. Li, and Y. Chen. 2017. PipeLayer: A Pipelined ReRAM-Based accelerator for deep learning. In 2017 IEEE International Symposium on High Performance Computer Architecture (HPCA). 541\u2013552."},{"key":"e_1_2_1_42_1","unstructured":"T. Tambe C. Hooper L. Pentecost E. Yang M. Donato V. Sanh A. M. Rush D. Brooks and G. Wei. 2020. EdgeBERT: Optimizing On-chip inference for multi-task NLP. arXiv preprint arXiv:2011.14203 (2020).  T. Tambe C. Hooper L. Pentecost E. Yang M. Donato V. Sanh A. M. Rush D. Brooks and G. Wei. 2020. EdgeBERT: Optimizing On-chip inference for multi-task NLP. arXiv preprint arXiv:2011.14203 (2020)."},{"key":"e_1_2_1_43_1","doi-asserted-by":"crossref","unstructured":"C. Torres-Huitzil and B. Girau. 2017. Fault and error tolerance in neural networks: A review. IEEE Access 5 (2017).  C. Torres-Huitzil and B. Girau. 2017. Fault and error tolerance in neural networks: A review. IEEE Access 5 (2017).","DOI":"10.1109\/ACCESS.2017.2742698"},{"key":"e_1_2_1_44_1","doi-asserted-by":"publisher","DOI":"10.1145\/3140659.3080244"},{"key":"e_1_2_1_45_1","unstructured":"M. Wang T. Xiao J. Li J. Zhang C. Hong and Z. Zhang. 2014. Minerva : A scalable and highly efficient training platform for deep learning.  M. Wang T. Xiao J. Li J. Zhang C. Hong and Z. Zhang. 2014. Minerva : A scalable and highly efficient training platform for deep learning."},{"key":"e_1_2_1_46_1","doi-asserted-by":"publisher","DOI":"10.1109\/JPROC.2012.2190369"},{"key":"e_1_2_1_47_1","doi-asserted-by":"publisher","DOI":"10.1109\/JETCAS.2017.2776980"},{"key":"e_1_2_1_48_1","doi-asserted-by":"publisher","DOI":"10.1145\/3061639.3062248"},{"key":"e_1_2_1_49_1","doi-asserted-by":"publisher","DOI":"10.1109\/TMAG.2005.855346"},{"key":"e_1_2_1_50_1","unstructured":"F. Yao A. S. Rakin and D. Fan. 2020. DeepHammer: Depleting the intelligence of deep neural networks through targeted chain of bit flips. (2020) 1463\u20131480.  F. Yao A. S. Rakin and D. Fan. 2020. DeepHammer: Depleting the intelligence of deep neural networks through targeted chain of bit flips. (2020) 1463\u20131480."},{"key":"e_1_2_1_51_1","doi-asserted-by":"publisher","DOI":"10.5555\/2014698.2014881"},{"key":"e_1_2_1_52_1","doi-asserted-by":"crossref","unstructured":"J. J. Zhang T. Gu K. Basu and S. Garg. 2018. Analyzing and mitigating the impact of permanent faults on a systolic array based neural network accelerator. (2018) 1\u20136.  J. J. Zhang T. Gu K. Basu and S. Garg. 2018. Analyzing and mitigating the impact of permanent faults on a systolic array based neural network accelerator. (2018) 1\u20136.","DOI":"10.1109\/VTS.2018.8368656"}],"container-title":["ACM Transactions on Embedded Computing Systems"],"original-title":[],"language":"en","link":[{"URL":"https:\/\/dl.acm.org\/doi\/10.1145\/3477016","content-type":"unspecified","content-version":"vor","intended-application":"text-mining"},{"URL":"https:\/\/dl.acm.org\/doi\/pdf\/10.1145\/3477016","content-type":"unspecified","content-version":"vor","intended-application":"similarity-checking"}],"deposited":{"date-parts":[[2025,6,17]],"date-time":"2025-06-17T19:30:46Z","timestamp":1750188646000},"score":1,"resource":{"primary":{"URL":"https:\/\/dl.acm.org\/doi\/10.1145\/3477016"}},"subtitle":[],"short-title":[],"issued":{"date-parts":[[2021,9,22]]},"references-count":52,"journal-issue":{"issue":"5s","published-print":{"date-parts":[[2021,10,31]]}},"alternative-id":["10.1145\/3477016"],"URL":"https:\/\/doi.org\/10.1145\/3477016","relation":{},"ISSN":["1539-9087","1558-3465"],"issn-type":[{"value":"1539-9087","type":"print"},{"value":"1558-3465","type":"electronic"}],"subject":[],"published":{"date-parts":[[2021,9,22]]},"assertion":[{"value":"2021-04-01","order":0,"name":"received","label":"Received","group":{"name":"publication_history","label":"Publication History"}},{"value":"2021-07-01","order":1,"name":"accepted","label":"Accepted","group":{"name":"publication_history","label":"Publication History"}},{"value":"2021-09-22","order":2,"name":"published","label":"Published","group":{"name":"publication_history","label":"Publication History"}}]}}