Ensemble Distribution Distillation for Self-Supervised Human Activity Recognition
This work addresses reliability and robustness issues in HAR for real-world applications, representing an incremental improvement by adapting existing methods to this domain.
The paper tackled challenges in Human Activity Recognition by applying Ensemble Distribution Distillation in a self-supervised framework, resulting in increased predictive accuracy, robust uncertainty estimates, and improved robustness against adversarial perturbations without added computational cost at inference.
Human Activity Recognition (HAR) has seen significant advancements with the adoption of deep learning techniques, yet challenges remain in terms of data requirements, reliability and robustness. This paper explores a novel application of Ensemble Distribution Distillation (EDD) within a self-supervised learning framework for HAR aimed at overcoming these challenges. By leveraging unlabeled data and a partially supervised training strategy, our approach yields an increase in predictive accuracy, robust estimates of uncertainty, and substantial increases in robustness against adversarial perturbation; thereby significantly improving reliability in real-world scenarios without increasing computational complexity at inference. We demonstrate this with an evaluation on several publicly available datasets. The contributions of this work include the development of a self-supervised EDD framework, an innovative data augmentation technique designed for HAR, and empirical validation of the proposed method's effectiveness in increasing robustness and reliability.