Enabling machine learning across heterogeneous sensor networks with graph autoencoders

Johan Medrano*, Fuchun Joseph Lin

*此作品的通信作者

研究成果: Conference contribution同行評審

2 引文 斯高帕斯(Scopus)

摘要

Machine Learning (ML) has been applied to enable many life-assisting applications, such as abnormality detection in daily routines and automatic emergency request for the solitary elderly. However, in most cases ML algorithms depend on the layout of the target Internet of Things (IoT) sensor network. Hence, to deploy an application across Heterogeneous Sensor Networks (HSNs), i.e. sensor networks with different sensors type or layouts, it is required to repeat the process of data collection and ML algorithm training. In this paper, we introduce a novel framework leveraging deep learning for graphs to enable using the same activity recognition system across HSNs deployed in different smart homes. Using our framework, we were able to transfer activity classifiers trained with activity labels on a source HSN to a target HSN, reaching about 75% of the baseline accuracy on the target HSN without using target activity labels. Moreover, our model can quickly adapt to unseen sensor layouts, which makes it highly suitable for the gradual deployment of real-world ML-based applications. In addition, we show that our framework is resilient to suboptimal graph representations of HSNs.

原文English
主出版物標題Ambient Intelligence - 15th European Conference, AmI 2019, Proceedings
編輯Ioannis Chatzigiannakis, Boris De Ruyter, Irene Mavrommati
發行者Springer
頁面153-169
頁數17
ISBN(列印)9783030342548
DOIs
出版狀態Published - 2019
事件15th European Conference on Ambient Intelligence, AmI 2019 - Rome, Italy
持續時間: 13 11月 201915 11月 2019

出版系列

名字Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
11912 LNCS
ISSN(列印)0302-9743
ISSN(電子)1611-3349

Conference

Conference15th European Conference on Ambient Intelligence, AmI 2019
國家/地區Italy
城市Rome
期間13/11/1915/11/19

指紋

深入研究「Enabling machine learning across heterogeneous sensor networks with graph autoencoders」主題。共同形成了獨特的指紋。

引用此