Abstract
Human Activity Recognition (HAR) from wearable sensors has gained significant attention in the last few decades, largely because of the potential healthcare benefits. For many years, HAR was done using classical machine learning approaches that require extraction of features. With the resurgence of deep learning, a major shift happened and at the moment, HAR researchers are mainly investigating different kinds of deep neural networks. However, deep learning comes with the challenge of having access to large amounts of labeled examples, which in the field of HAR is considered an expensive task, both in terms of time and effort. Another challenge is the fact that the training and testing data in HAR can be different due to the personal preferences of different people when performing the same activity. In order to try and mitigate these problems, in this paper we explore transfer learning, a paradigm for transferring knowledge from a source domain, to another related target domain. More specifically, we explore the effects of transferring knowledge between two open-source datasets, the Opportunity and JSI-FOS datasets, using weight-transfer for the DeepConvLSTM architecture. We also explore the performance of this transfer at different amounts of labeled data from the target domain. The experiments showed that it is beneficial to transfer the weights of fewer layers, and that deep transfer learning can perform better than a domain-specific deep end-to-end model in specific circumstances. Finally, we show that deep transfer learning is a viable alternative to classical machine learning approaches as it produces comparable results and does not require feature extraction.
Highlights
With numerous healthcare, smart-home and security applications on the horizon, human activity recognition (HAR) is a field which has gained significant traction in the past two decades
We compare the performance of a Random Forest (RF) classifier, an end-to-end (E2E) DeepConvLSTM model and a DeepConvLSTM model trained using transfer learning
It is interesting to note that it comes close, the E2E model never really matches the performance of the Random Forest (RF) classifier
Summary
Smart-home and security applications on the horizon, human activity recognition (HAR) is a field which has gained significant traction in the past two decades. Most of the research done on this topic has been aimed at understanding human activity using data from wearable sensors, primarily inertial measurement units (IMUs). This is in large part due to the rapid development of wearable devices, allowing researchers to perform complex tasks on them, as well as their ubiquitous and unobtrusive nature. Researchers in the field of HAR were mainly focused on using traditional pattern-matching methods as well as machine learning to detect activities [1]. The focus has mostly been centered around Convolutional Neural Networks (CNNs).
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have