On-device training of machine learning models on microcontrollers with a look at federated learning
Document typeConference report
PublisherAssociation for Computing Machinery (ACM)
Rights accessOpen Access
Recent progress in machine learning frameworks makes it now possible to run an inference with sophisticated machine learning models on tiny microcontrollers. Model training, however, is typically done separately on powerful computers. There, the training process has abundant CPU and memory resources to process the stored datasets. In this work, we explore a different approach: training the model directly on the microcontroller. We implement this approach for a keyword spotting task. Then, we extend the training process using federated learning among microcontrollers. Our experiments with model training show an overall trend of decreasing loss with the increase of training epochs.
CitationMonfort, M.; Pueyo, R.; Freitag, F. On-device training of machine learning models on microcontrollers with a look at federated learning. A: ACM International Conference on Information Technology for Social Good. "GoodIT'21: proceedings of the 2021 Conference on Information Technology for Social Good: September 9–11, 2021, Roma, Italy". New York: Association for Computing Machinery (ACM), 2021, p. 198-203. ISBN 978-1-4503-8478-0. DOI 10.1145/3462203.3475896.
All rights reserved. This work is protected by the corresponding intellectual and industrial property rights. Without prejudice to any existing legal exemptions, reproduction, distribution, public communication or transformation of this work are prohibited without permission of the copyright holder