A Novel Supervised Heterogenuos Feature Transfer Learning Scheme for ADL Recognition
Batsergelen Myagmar, Jie Li, Shigetomo Kimura · 2018
In many of real-world applications of machine learning, one of the leading issues is the high cost of obtaining labeled data for learning tasks. One promising research area for solving this problem is transfer learning, which aims to learn a task in target domain by utilizing the training data in a different but related source domain. In this paper, we propose a novel supervised transfer learning algorithm that discovers good feature representations across different domains by deriving domain invariant feature space, where data distributions in different domains are close to one another. Within the new feature space, we apply classification algorithm to train classifiers in the source domain for use in the target domain. We verify the effectiveness of our approach by experimenting on real-world smart home datasets and shows that our approach extends the accuracy of transfer learning for activities of daily living (ADL) recognition.