TY - GEN
T1 - Personalized Sleep State Classification via Learned Factor Graphs
AU - Rubinstein, Bar
AU - Filin, Yoav
AU - Shlezinger, Nir
AU - Farsad, Nariman
N1 - DBLP License: DBLP's bibliographic metadata records provided through http://dblp.org/ are distributed under a Creative Commons CC0 1.0 Universal Public Domain Dedication. Although the bibliographic metadata records are provided consistent with CC0 1.0 Dedication, the content described by the metadata records is not. Content may be subject to copyright, rights of privacy, rights of publicity and other restrictions.
PY - 2022/10/18
Y1 - 2022/10/18
N2 - Recent years have witnessed a growing interest in using deep learning for sleep state tracking. A key challenge in doing so stems from the variability between different patients, and a model trained using the data of several patients may perform poorly on another subject. In this work, we study mechanisms for achieving personalized sleep state tracking, which can be utilized with various deep neural network (DNN) architectures. Our design uses learned factor graphs to exploit temporal correlation in a principled manner. Inspired by recent advances in federated learning, we incorporate schemes based on data and model interpolation for achieving personalized models. Our experimental study demonstrates that this approach achieves accurate classification with compact DNNs.
AB - Recent years have witnessed a growing interest in using deep learning for sleep state tracking. A key challenge in doing so stems from the variability between different patients, and a model trained using the data of several patients may perform poorly on another subject. In this work, we study mechanisms for achieving personalized sleep state tracking, which can be utilized with various deep neural network (DNN) architectures. Our design uses learned factor graphs to exploit temporal correlation in a principled manner. Inspired by recent advances in federated learning, we incorporate schemes based on data and model interpolation for achieving personalized models. Our experimental study demonstrates that this approach achieves accurate classification with compact DNNs.
KW - Sleep state monitoring
KW - deep learning
KW - signal processing
UR - http://www.scopus.com/inward/record.url?scp=85141010936&partnerID=8YFLogxK
U2 - 10.23919/EUSIPCO55093.2022.9909523
DO - 10.23919/EUSIPCO55093.2022.9909523
M3 - Conference contribution
T3 - European Signal Processing Conference
SP - 1427
EP - 1431
BT - 30th European Signal Processing Conference, EUSIPCO 2022 - Proceedings
PB - European Signal Processing Conference, EUSIPCO
T2 - 30th European Signal Processing Conference, EUSIPCO 2022
Y2 - 29 August 2022 through 2 September 2022
ER -