Wearable-based behaviour interpolation for semi-supervised human activity recognition
This work addresses the problem of data labeling inefficiency for researchers and practitioners in wearable computing and activity recognition, presenting an incremental advancement in semi-supervised methods for HAR.
The paper tackles the challenge of requiring large labeled datasets for deep learning-based Human Activity Recognition (HAR) by introducing MixHAR, a semi-supervised approach that blends labeled and unlabeled activities using linear interpolation and a mixing calibration mechanism to address activity intrusion, resulting in significant performance improvements.
While traditional feature engineering for Human Activity Recognition (HAR) involves a trial-anderror process, deep learning has emerged as a preferred method for high-level representations of sensor-based human activities. However, most deep learning-based HAR requires a large amount of labelled data and extracting HAR features from unlabelled data for effective deep learning training remains challenging. We, therefore, introduce a deep semi-supervised HAR approach, MixHAR, which concurrently uses labelled and unlabelled activities. Our MixHAR employs a linear interpolation mechanism to blend labelled and unlabelled activities while addressing both inter- and intra-activity variability. A unique challenge identified is the activityintrusion problem during mixing, for which we propose a mixing calibration mechanism to mitigate it in the feature embedding space. Additionally, we rigorously explored and evaluated the five conventional/popular deep semi-supervised technologies on HAR, acting as the benchmark of deep semi-supervised HAR. Our results demonstrate that MixHAR significantly improves performance, underscoring the potential of deep semi-supervised techniques in HAR.