Learning personalized ADL recognition models from few raw data
作者：
Highlights：
• A new neural architecture, combining matching networks with sequence to sequence models, called SSMN.
• An SSMN application to few shot inertial sequence training for personalized activities of daily living recognition.
• Fine-tuning the SSMN neural architecture by using a pretrain model on intertial posture data.
• High performances in activities of daily living recognition allowing robust actigraphy system that estimates elderly people autonomy.
摘要
•A new neural architecture, combining matching networks with sequence to sequence models, called SSMN.•An SSMN application to few shot inertial sequence training for personalized activities of daily living recognition.•Fine-tuning the SSMN neural architecture by using a pretrain model on intertial posture data.•High performances in activities of daily living recognition allowing robust actigraphy system that estimates elderly people autonomy.
论文关键词：Few-shot learning,Matching networks,Activity of daily living,EHealth,Inertial measurement unit,Gated recurrent units
论文评审过程：Received 18 November 2019, Revised 25 April 2020, Accepted 23 June 2020, Available online 27 June 2020, Version of Record 2 July 2020.
论文官网地址：https://doi.org/10.1016/j.artmed.2020.101916
