Cooperative Random Forest for Privacy-Preserving IoT Devices
Yui Yamashita, Akihito Taya, Yoshito Tobe · 2020
Recently, various Internet of things (IoT) devices have become widely used in our daily lives and made houses and cities easier to live in. This paper proposes a machine learning scheme to take advantage of IoT devices. The proposed scheme realizes cooperation between devices to improve their performance, rather than learning independently. However, it is difficult to share local data directly because those data may contain private information, such as a picture with a user's face or lifelog data. Therefore, this paper provides a way of preserving privacy in interconnected IoT devices by sharing only learners from each device without sharing the original data directly. The proposed algorithm shares decision trees locally learned at each device and utilizes a random forest as a way of combining them together.