{"status":"ok","message-type":"work","message-version":"1.0.0","message":{"indexed":{"date-parts":[[2026,4,28]],"date-time":"2026-04-28T12:30:17Z","timestamp":1777379417119,"version":"3.51.4"},"reference-count":43,"publisher":"Association for Computing Machinery (ACM)","issue":"2","license":[{"start":{"date-parts":[[2019,6,21]],"date-time":"2019-06-21T00:00:00Z","timestamp":1561075200000},"content-version":"vor","delay-in-days":0,"URL":"https:\/\/www.acm.org\/publications\/policies\/copyright_policy#Background"}],"content-domain":{"domain":["dl.acm.org"],"crossmark-restriction":true},"short-container-title":["Proc. ACM Interact. Mob. Wearable Ubiquitous Technol."],"published-print":{"date-parts":[[2019,6,21]]},"abstract":"<jats:p>There is growing interest in being able to run neural networks on sensors, wearables and internet-of-things (IoT) devices. However, the computational demands of neural networks make them difficult to deploy on resource-constrained edge devices.<\/jats:p>\n          <jats:p>To meet this need, our work introduces a new recurrent unit architecture that is specifically adapted for on-device low power acoustic event detection (AED). The proposed architecture is based on the gated recurrent unit ('GRU' -- introduced by Cho et al. [9]) but features optimizations that make it implementable on ultra-low power micro-controllers such as the Arm Cortex M0+.<\/jats:p>\n          <jats:p>Our new architecture, the Embedded Gated Recurrent Unit (eGRU) is demonstrated to be highly efficient and suitable for short-duration AED and keyword spotting tasks. A single eGRU cell is 60\u00d7 faster and 10\u00d7 smaller than a GRU cell. Despite its optimizations, eGRU compares well with GRU across tasks of varying complexities.<\/jats:p>\n          <jats:p>The practicality of eGRU is investigated in a wearable acoustic event detection application. An eGRU model is implemented and tested on the Arm Cortex M0-based Atmel ATSAMD21E18 processor. The Arm M0+ implementation of the eGRU model compares favorably with a full precision GRU that is running on a workstation. The embedded eGRU model achieves a classification accuracy 95.3%, which is only 2% less than the full precision GRU.<\/jats:p>","DOI":"10.1145\/3328907","type":"journal-article","created":{"date-parts":[[2019,6,24]],"date-time":"2019-06-24T13:45:01Z","timestamp":1561383901000},"page":"1-17","update-policy":"https:\/\/doi.org\/10.1145\/crossmark-policy","source":"Crossref","is-referenced-by-count":18,"title":["An Optimized Recurrent Unit for Ultra-Low-Power Keyword Spotting"],"prefix":"10.1145","volume":"3","author":[{"given":"Justice","family":"Amoh","sequence":"first","affiliation":[{"name":"Thayer School of Engineering, Dartmouth College, Hanover, NH, USA"}],"role":[{"role":"author","vocabulary":"crossref"}]},{"given":"Kofi M.","family":"Odame","sequence":"additional","affiliation":[{"name":"Thayer School of Engineering, Dartmouth College, Hanover, NH, USA"}],"role":[{"role":"author","vocabulary":"crossref"}]}],"member":"320","published-online":{"date-parts":[[2019,6,21]]},"reference":[{"key":"e_1_2_2_1_1","unstructured":"{n. d.}. Autonomous Car Development Platform from NVIDIA DRIVE PX2. https:\/\/www.nvidia.com\/en-au\/self-driving-cars\/drive-px\/ Accessed: 2018-09-17.  {n. d.}. Autonomous Car Development Platform from NVIDIA DRIVE PX2. https:\/\/www.nvidia.com\/en-au\/self-driving-cars\/drive-px\/ Accessed: 2018-09-17."},{"key":"e_1_2_2_2_1","unstructured":"2017. iPhone 8 and iPhone 8 Plus: A new generation of iPhone. https:\/\/www.apple.com\/newsroom\/2017\/09\/iphone-8-and-iphone-8-plus-a-new-generation-of-iphone\/ Accessed: 2018-09-17.  2017. iPhone 8 and iPhone 8 Plus: A new generation of iPhone. https:\/\/www.apple.com\/newsroom\/2017\/09\/iphone-8-and-iphone-8-plus-a-new-generation-of-iphone\/ Accessed: 2018-09-17."},{"key":"e_1_2_2_3_1","first-page":"265","article-title":"Tensorflow: a system for large-scale machine learning","volume":"16","author":"Abadi Mart\u00edn","year":"2016","journal-title":"OSDI"},{"key":"e_1_2_2_4_1","unstructured":"Justice Amoh and Kofi Odame. 2013. Technologies for developing ambulatory cough monitoring devices. Critical Reviews\u2122 in Biomedical Engineering 41 6 (2013).  Justice Amoh and Kofi Odame. 2013. Technologies for developing ambulatory cough monitoring devices. Critical Reviews\u2122 in Biomedical Engineering 41 6 (2013)."},{"key":"e_1_2_2_5_1","doi-asserted-by":"publisher","DOI":"10.1109\/BioCAS.2015.7348395"},{"key":"e_1_2_2_6_1","doi-asserted-by":"crossref","unstructured":"Justice Amoh and Kofi Odame. 2016. Deep neural networks for identifying cough sounds. IEEE transactions on biomedical circuits and systems 10 5 (2016) 1003--1011.  Justice Amoh and Kofi Odame. 2016. Deep neural networks for identifying cough sounds. IEEE transactions on biomedical circuits and systems 10 5 (2016) 1003--1011.","DOI":"10.1109\/TBCAS.2016.2598794"},{"key":"e_1_2_2_7_1","volume-title":"Mobile System For Self-Monitoring Of Asthma. In Engineering in Medicine and Biology Conference (EMBC)","author":"Amoh Justice","year":"2014"},{"key":"e_1_2_2_8_1","doi-asserted-by":"publisher","DOI":"10.1109\/MICRO.2014.58"},{"key":"e_1_2_2_9_1","doi-asserted-by":"publisher","DOI":"10.3115\/v1\/D14-1179"},{"key":"e_1_2_2_10_1","unstructured":"Junyoung Chung Caglar Gulcehre KyungHyun Cho and Yoshua Bengio. 2014. Empirical evaluation of gated recurrent neural networks on sequence modeling. arXiv preprint arXiv:1412.3555 (2014).  Junyoung Chung Caglar Gulcehre KyungHyun Cho and Yoshua Bengio. 2014. Empirical evaluation of gated recurrent neural networks on sequence modeling. arXiv preprint arXiv:1412.3555 (2014)."},{"key":"e_1_2_2_11_1","unstructured":"Jasmine Collins Jascha Sohl-Dickstein and David Sussillo. 2016. Capacity and trainability in recurrent neural networks. arXiv preprint arXiv:1611.09913 (2016).  Jasmine Collins Jascha Sohl-Dickstein and David Sussillo. 2016. Capacity and trainability in recurrent neural networks. arXiv preprint arXiv:1611.09913 (2016)."},{"key":"e_1_2_2_12_1","unstructured":"Atmel Corporation. 2015. ATSAMD21E Datasheet. https:\/\/cdn.sparkfun.com\/datasheets\/Dev\/Arduino\/Boards\/Atmel-42181-SAM-D21_Datasheet.pdf  Atmel Corporation. 2015. ATSAMD21E Datasheet. https:\/\/cdn.sparkfun.com\/datasheets\/Dev\/Arduino\/Boards\/Atmel-42181-SAM-D21_Datasheet.pdf"},{"key":"e_1_2_2_13_1","first-page":"1","article-title":"Training deep neural networks with low precision multiplications","volume":"5","author":"Courbariaux Matthieu","year":"2014","journal-title":"Section"},{"key":"e_1_2_2_14_1","doi-asserted-by":"publisher","DOI":"10.1109\/AINA.2010.187"},{"key":"e_1_2_2_15_1","volume-title":"Proceedings of the thirteenth international conference on artificial intelligence and statistics. 249--256","author":"Glorot Xavier","year":"2010"},{"key":"e_1_2_2_16_1","doi-asserted-by":"publisher","DOI":"10.1145\/1143844.1143891"},{"key":"e_1_2_2_17_1","unstructured":"Suyog Gupta Ankur Agrawal Kailash Gopalakrishnan and Pritish Narayanan. 2015. Deep Learning with Limited Numerical Precision. 37 (2015). arXiv:1502.02551  Suyog Gupta Ankur Agrawal Kailash Gopalakrishnan and Pritish Narayanan. 2015. Deep Learning with Limited Numerical Precision. 37 (2015). arXiv:1502.02551"},{"key":"e_1_2_2_18_1","unstructured":"Song Han Huizi Mao and William J. Dally. 2015. Deep Compression: Compressing Deep Neural Networks with Pruning Trained Quantization and Huffman Coding. (2015) 1--14. https:\/\/doi.org\/abs\/1510.00149\/1510.00149 arXiv:1510.00149  Song Han Huizi Mao and William J. Dally. 2015. Deep Compression: Compressing Deep Neural Networks with Pruning Trained Quantization and Huffman Coding. (2015) 1--14. https:\/\/doi.org\/abs\/1510.00149\/1510.00149 arXiv:1510.00149"},{"key":"e_1_2_2_19_1","unstructured":"Awni Hannun Carl Case Jared Casper Bryan Catanzaro Greg Diamos Erich Elsen Ryan Prenger Sanjeev Satheesh Shubho Sengupta Adam Coates etal 2014. Deep speech: Scaling up end-to-end speech recognition. arXiv preprint arXiv:1412.5567 (2014).  Awni Hannun Carl Case Jared Casper Bryan Catanzaro Greg Diamos Erich Elsen Ryan Prenger Sanjeev Satheesh Shubho Sengupta Adam Coates et al. 2014. Deep speech: Scaling up end-to-end speech recognition. arXiv preprint arXiv:1412.5567 (2014)."},{"key":"e_1_2_2_20_1","doi-asserted-by":"publisher","DOI":"10.1162\/neco.1997.9.8.1735"},{"key":"e_1_2_2_21_1","volume-title":"Mobilenets: Efficient convolutional neural networks for mobile vision applications. arXiv preprint arXiv:1704.04861","author":"Howard Andrew G","year":"2017"},{"key":"e_1_2_2_22_1","unstructured":"Itay Hubara Matthieu Courbariaux Daniel Soudry Ran El-Yaniv and Yoshua Bengio. 2016. Quantized Neural Networks: Training Neural Networks with Low Precision Weights and Activations. (2016) 1--29. arXiv:1609.07061 http:\/\/arxiv.org\/abs\/1609.07061   Itay Hubara Matthieu Courbariaux Daniel Soudry Ran El-Yaniv and Yoshua Bengio. 2016. Quantized Neural Networks: Training Neural Networks with Low Precision Weights and Activations. (2016) 1--29. arXiv:1609.07061 http:\/\/arxiv.org\/abs\/1609.07061"},{"key":"e_1_2_2_23_1","unstructured":"Liangzhen Lai Naveen Suda and Vikas Chandra. 2018. CMSIS-NN: Efficient Neural Network Kernels for Arm Cortex-M CPUs. (2018) 1--10. arXiv:1801.06601 http:\/\/arxiv.org\/abs\/1801.06601  Liangzhen Lai Naveen Suda and Vikas Chandra. 2018. CMSIS-NN: Efficient Neural Network Kernels for Arm Cortex-M CPUs. (2018) 1--10. arXiv:1801.06601 http:\/\/arxiv.org\/abs\/1801.06601"},{"key":"e_1_2_2_24_1","unstructured":"Arm Ltd. {n. d.}. Cortex-M0 -- Arm Developer. https:\/\/developer.arm.com\/products\/processors\/cortex-m\/cortex-m0-plus Accessed: 2010-09-30.  Arm Ltd. {n. d.}. Cortex-M0 -- Arm Developer. https:\/\/developer.arm.com\/products\/processors\/cortex-m\/cortex-m0-plus Accessed: 2010-09-30."},{"key":"e_1_2_2_25_1","unstructured":"Paul Merolla Rathinakumar Appuswamy John Arthur Steve K Esser and Dharmendra Modha. 2016. Deep neural networks are robust to weight binarization and other non-linear distortions. arXiv preprint arXiv:1606.01981 (2016).  Paul Merolla Rathinakumar Appuswamy John Arthur Steve K Esser and Dharmendra Modha. 2016. Deep neural networks are robust to weight binarization and other non-linear distortions. arXiv preprint arXiv:1606.01981 (2016)."},{"key":"e_1_2_2_26_1","unstructured":"Microchip Technology 2015. 20-Pin Flash 8-Bit Microcontrollers with XLP Technology. Microchip Technology. Rev. 1.0.  Microchip Technology 2015. 20-Pin Flash 8-Bit Microcontrollers with XLP Technology. Microchip Technology. Rev. 1.0."},{"key":"e_1_2_2_27_1","first-page":"4","article-title":"Energy Efficiency of Mobile Clients in Cloud Computing","volume":"10","author":"Miettinen Antti P","year":"2010","journal-title":"HotCloud"},{"key":"e_1_2_2_28_1","unstructured":"Joachim Ott Zhouhan Lin Ying Zhang Shih-Chii Liu and Yoshua Bengio. 2016. Recurrent Neural Networks With Limited Numerical Precision. (2016) 1--11. arXiv:1611.07065 http:\/\/arxiv.org\/abs\/1611.07065  Joachim Ott Zhouhan Lin Ying Zhang Shih-Chii Liu and Yoshua Bengio. 2016. Recurrent Neural Networks With Limited Numerical Precision. (2016) 1--11. arXiv:1611.07065 http:\/\/arxiv.org\/abs\/1611.07065"},{"key":"e_1_2_2_29_1","volume-title":"International Conference on Machine Learning. 1310--1318","author":"Pascanu Razvan","year":"2013"},{"key":"e_1_2_2_30_1","unstructured":"Adam Paszke Sam Gross Soumith Chintala Gregory Chanan Edward Yang Zachary DeVito Zeming Lin Alban Desmaison Luca Antiga and Adam Lerer. 2017. Automatic differentiation in PyTorch. In NIPS-W.  Adam Paszke Sam Gross Soumith Chintala Gregory Chanan Edward Yang Zachary DeVito Zeming Lin Alban Desmaison Luca Antiga and Adam Lerer. 2017. Automatic differentiation in PyTorch. In NIPS-W."},{"key":"e_1_2_2_31_1","doi-asserted-by":"publisher","DOI":"10.21437\/Interspeech.2017-775"},{"key":"e_1_2_2_32_1","doi-asserted-by":"crossref","unstructured":"Daniele Rav\u0131 Charence Wong Fani Deligianni Melissa Berthelot Javier Andreu-Perez Benny Lo and Guang-Zhong Yang. 2017. Deep learning for health informatics. IEEE journal of biomedical and health informatics 21 1 (2017) 4--21.  Daniele Rav\u0131 Charence Wong Fani Deligianni Melissa Berthelot Javier Andreu-Perez Benny Lo and Guang-Zhong Yang. 2017. Deep learning for health informatics. IEEE journal of biomedical and health informatics 21 1 (2017) 4--21.","DOI":"10.1109\/JBHI.2016.2636665"},{"key":"e_1_2_2_33_1","doi-asserted-by":"publisher","DOI":"10.1145\/2647868.2655045"},{"issue":"0","key":"e_1_2_2_34_1","article-title":"Ultra Low Power 128K, LCD MCU Family","volume":"1","author":"Labs Silicon","year":"2013","journal-title":"Silicon Labs. Rev."},{"key":"e_1_2_2_35_1","unstructured":"STMicroelectronics 2018. Ultra-low-power value line Arm Cortex-M0+ MCU. STMicroelectronics. Rev. 2.0.  STMicroelectronics 2018. Ultra-low-power value line Arm Cortex-M0+ MCU. STMicroelectronics. Rev. 2.0."},{"key":"e_1_2_2_36_1","unstructured":"Ilya Sutskever Oriol Vinyals and Quoc V Le. 2014. Sequence to sequence learning with neural networks. In Advances in neural information processing systems. 3104--3112.   Ilya Sutskever Oriol Vinyals and Quoc V Le. 2014. Sequence to sequence learning with neural networks. In Advances in neural information processing systems. 3104--3112."},{"key":"e_1_2_2_37_1","doi-asserted-by":"publisher","DOI":"10.1109\/MSP.2010.186"},{"key":"e_1_2_2_38_1","unstructured":"Neil Tan. 2017. uTensor: AI Inference library based on mbed and TensorFlow. \\url{https:\/\/github.com\/uTensor\/uTensor}.  Neil Tan. 2017. uTensor: AI Inference library based on mbed and TensorFlow. \\url{https:\/\/github.com\/uTensor\/uTensor}."},{"key":"e_1_2_2_39_1","doi-asserted-by":"publisher","DOI":"10.1109\/ICDCS.2017.226"},{"key":"e_1_2_2_40_1","article-title":"Ultra-Low-Power Mixed-Signal Microcontrollers. Texas","author":"Instruments Texas","year":"2018","journal-title":"Instruments. Rev. N."},{"key":"e_1_2_2_41_1","volume-title":"Speech Commands: A Dataset for Limited-Vocabulary Speech Recognition.","author":"Warden Pete","year":"2018"},{"key":"e_1_2_2_42_1","volume-title":"Quantized Convolutional Neural Networks for Mobile Devices. 2016 IEEE Conference on Computer Vision and Pattern Recognition (CVPR)","author":"Wu Jiaxiang","year":"2016"},{"key":"e_1_2_2_43_1","doi-asserted-by":"publisher","DOI":"10.1007\/s11633-016-1006-2"}],"container-title":["Proceedings of the ACM on Interactive, Mobile, Wearable and Ubiquitous Technologies"],"original-title":[],"language":"en","link":[{"URL":"https:\/\/dl.acm.org\/doi\/10.1145\/3328907","content-type":"unspecified","content-version":"vor","intended-application":"text-mining"},{"URL":"https:\/\/dl.acm.org\/doi\/pdf\/10.1145\/3328907","content-type":"unspecified","content-version":"vor","intended-application":"similarity-checking"}],"deposited":{"date-parts":[[2025,6,17]],"date-time":"2025-06-17T23:54:40Z","timestamp":1750204480000},"score":1,"resource":{"primary":{"URL":"https:\/\/dl.acm.org\/doi\/10.1145\/3328907"}},"subtitle":[],"short-title":[],"issued":{"date-parts":[[2019,6,21]]},"references-count":43,"journal-issue":{"issue":"2","published-print":{"date-parts":[[2019,6,21]]}},"alternative-id":["10.1145\/3328907"],"URL":"https:\/\/doi.org\/10.1145\/3328907","relation":{},"ISSN":["2474-9567"],"issn-type":[{"value":"2474-9567","type":"electronic"}],"subject":[],"published":{"date-parts":[[2019,6,21]]},"assertion":[{"value":"2018-11-01","order":0,"name":"received","label":"Received","group":{"name":"publication_history","label":"Publication History"}},{"value":"2019-04-01","order":1,"name":"accepted","label":"Accepted","group":{"name":"publication_history","label":"Publication History"}},{"value":"2019-06-21","order":2,"name":"published","label":"Published","group":{"name":"publication_history","label":"Publication History"}}]}}