Sufficient physical activity and restful sleep play a major role in the prevention and cure of many chronic conditions. Being able to proactively screen and monitor such chronic conditions would be a big step forward for overall health. The rapid increase in the popularity of wearable devices provides a significant new source, making it possible to track the user’s lifestyle real-time. In this paper, we propose a novel unsupervised representation learning technique called activ- ity2vec that learns and “summarizes” the discrete-valued ac- tivity time-series. It learns the representations with three com- ponents: (i) the co-occurrence and magnitude of the activity levels in a time-segment, (ii) neighboring context of the time- segment, and (iii) promoting subject-invariance with adver- sarial training. We evaluate our method on four disorder pre- diction tasks using linear classifiers. Empirical evaluation and analysis demonstrate that our proposed method performs bet- ter than many strong baselines, and adversarial learning helps improve the generalizability of our representations by pro- moting subject invariant features. We also show that using the representations at the level of a day works the best since human activity is structured in terms of daily routines.

@inproceedings{karan-et-al-aaai-19, abstract = {Sufficient physical activity and restful sleep play a major role in the prevention and cure of many chronic conditions. Being able to proactively screen and monitor such chronic conditions would be a big step forward for overall health. The rapid increase in the popularity of wearable devices provides a significant new source, making it possible to track the user’s lifestyle real-time. In this paper, we propose a novel unsupervised representation learning technique called activ- ity2vec that learns and “summarizes” the discrete-valued ac- tivity time-series. It learns the representations with three com- ponents: (i) the co-occurrence and magnitude of the activity levels in a time-segment, (ii) neighboring context of the time- segment, and (iii) promoting subject-invariance with adver- sarial training. We evaluate our method on four disorder pre- diction tasks using linear classifiers. Empirical evaluation and analysis demonstrate that our proposed method performs bet- ter than many strong baselines, and adversarial learning helps improve the generalizability of our representations by pro- moting subject invariant features. We also show that using the representations at the level of a day works the best since human activity is structured in terms of daily routines.}, address = {Honolulu, Hawaii}, author = {Karan Aggarwal and Shafiq Joty and Luis Fernandez-Luque and Jaideep Srivastava}, booktitle = {Thirty-Third AAAI Conference on Artificial Intelligence}, month = {September}, pages = {834 -- 841}, publisher = {AAAI}, series = {AAAI'19}, title = {Adversarial Unsupervised Representation Learning for Activity Time-Series}, url = {https://arxiv.org/abs/1811.06847}, year = {2019} }