| 大会名称 |
| 2019年 総合大会 |
| 大会コ－ド |
| 2019G |
| 開催年 |
|
2019
|
| 発行日 |
|
2019-03-05
|
| セッション番号 |
|
D-9
|
| セッション名 |
|
ライフインテリジェンスとオフィス情報システム
|
| 講演日 |
|
2019/03/19
|
| 講演場所(会議室等) |
|
54号館 401教室
|
| 講演番号 |
|
D-9-4
|
| タイトル |
|
Deep Transfer Learning of Indoor Human Activity Recognition Across Households
|
| 著者名 |
|
○Hao Niu, Kei Yonekawa, Mori Kurokawa, Shinya Wada, Kiyohito Yoshihara,
|
| キーワード |
|
Indoor Human Activity Recognition, Transfer Learning, Generative Adversarial Network
|
| 抄録 |
|
Human activity recognition (HAR) using sensor data has been studied extensively. Generally, HAR is done individually for each domain (e.g., household). However, in some cases the data of some domains cannot be labelled due to the practical or privacy problems. The solution may be directly reusing the model built for other domains or adopting transfer learning techniques. In this paper, we collect the real sensor data of 3 households and evaluate the performance of applying an existing GAN-based transfer learning approach to the indoor HAR across these households.
|
| 本文pdf |
|
PDF download
|
