SOS: A Shuffle Order Strategy for Data Augmentation in Industrial Human Activity Recognition
This addresses data scarcity and heterogeneity issues in industrial HAR systems, offering incremental improvements for real-world applications.
The study tackled the challenge of obtaining high-quality and varied data for Human Activity Recognition (HAR) by introducing a dataset generation method using deep learning and a shuffle order strategy to homogenize data distribution, resulting in improved classification performance with an accuracy of up to 0.70 ± 0.03 and a macro F1 score of 0.64 ± 0.01.
In the realm of Human Activity Recognition (HAR), obtaining high quality and variance data is still a persistent challenge due to high costs and the inherent variability of real-world activities. This study introduces a generation dataset by deep learning approaches (Attention Autoencoder and conditional Generative Adversarial Networks). Another problem that data heterogeneity is a critical challenge, one of the solutions is to shuffle the data to homogenize the distribution. Experimental results demonstrate that the random sequence strategy significantly improves classification performance, achieving an accuracy of up to 0.70 $\pm$ 0.03 and a macro F1 score of 0.64 $\pm$ 0.01. For that, disrupting temporal dependencies through random sequence reordering compels the model to focus on instantaneous recognition, thereby improving robustness against activity transitions. This approach not only broadens the effective training dataset but also offers promising avenues for enhancing HAR systems in complex, real-world scenarios.