TY - CPAPER
KW - embedded systems
KW - federated learning
KW - keyword spotting
KW - machine learning
AU - Marc Grau
AU - Roger Pueyo
AU - Felix Freitag
AB - 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.
CY - New York, NY, USA
DO - 10.1145/3462203.3475896
N2 - 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.
PB - Association for Computing Machinery
PP - New York, NY, USA
PY - 2021
SN - 9781450384780
EP - 198–203
TI - On-Device Training of Machine Learning Models on Microcontrollers With a Look at Federated Learning
UR - https://doi.org/10.1145/3462203.3475896
ER -