مؤتمر
Mixed-precision quantization and parallel implementation of multispectral Riemannian classification for brain-machine interfaces
العنوان: | Mixed-precision quantization and parallel implementation of multispectral Riemannian classification for brain-machine interfaces |
---|---|
المؤلفون: | Wang X., Schneider T., Hersche M., Cavigelli L., Benini L. |
المساهمون: | Wang X., Schneider T., Hersche M., Cavigelli L., Benini L. |
بيانات النشر: | Institute of Electrical and Electronics Engineers Inc. 345 E 47TH ST, NEW YORK, NY 10017 USA |
سنة النشر: | 2021 |
المجموعة: | IRIS Università degli Studi di Bologna (CRIS - Current Research Information System) |
مصطلحات موضوعية: | Brain-machine interface, Deep learning, Edge computing, Machine learning, Motor imagery, Parallel computing |
الوصف: | With Motor-Imagery (MI) Brain-Machine Interfaces (BMIs) we may control machines by merely thinking of performing a motor action. Practical use cases require a wearable solution where the classification of the brain signals is done locally near the sensor using machine learning models embedded on energy-efficient microcontroller units (MCUs), for assured privacy, user comfort, and long-term usage. In this work, we provide practical insights on the accuracy-cost tradeoff for embedded BMI solutions. Our proposed Multispectral Riemannian Classifier reaches 75.1% accuracy on 4-class MI task. We further scale down the model by quantizing it to mixed-precision representations with a minimal accuracy loss of 1%, which is still 3.2% more accurate than the state-of-the-art embedded convolutional neural network. We implement the model on a low-power MCU with parallel processing units taking only 33.39 ms and consuming 1.304 mJ per classification. |
نوع الوثيقة: | conference object |
وصف الملف: | ELETTRONICO |
اللغة: | English |
العلاقة: | info:eu-repo/semantics/altIdentifier/isbn/978-1-7281-9201-7; info:eu-repo/semantics/altIdentifier/wos/WOS:000706507900087; ispartofbook:Proceedings - IEEE International Symposium on Circuits and Systems; 53rd IEEE International Symposium on Circuits and Systems, ISCAS 2021; volume:2021-; firstpage:1; lastpage:5; numberofpages:5; http://hdl.handle.net/11585/869392Test; info:eu-repo/semantics/altIdentifier/scopus/2-s2.0-85109010829 |
DOI: | 10.1109/ISCAS51556.2021.9401564 |
الإتاحة: | https://doi.org/10.1109/ISCAS51556.2021.9401564Test http://hdl.handle.net/11585/869392Test |
رقم الانضمام: | edsbas.9E96F0D7 |
قاعدة البيانات: | BASE |
DOI: | 10.1109/ISCAS51556.2021.9401564 |
---|