STARS: Soft Multi-Task Learning for Activity Recognition from Multi-Modal Sensor Data
Human activity recognition from ubiquitous sensor data is an important but challenging classification problem for applications such as assisted living, energy management, and security monitoring of smart homes. In this paper, we present a soft probabilistic classification model for human activity recognition from multi-modal sensors in a smart home environment. The model employs a softmax multi-task learning approach to fit a joint model for all the rooms in the smart home, taking into account the diverse types of sensors available in different rooms. The model also learns the transitional dependencies between activities to improve its prediction accuracy. Experimental results on a real-world dataset showed that the proposed approach outperforms several baseline methods, including k-nearest neighbors, conditional random field, and standard multinomial logistic regression.
This research is supported in part by the U.S. National Science Foundation through grant NSF III-1615612. Any use of trade, firm, or product names is for descriptive purposes only and does not imply endorsement by the U.S. Government.
- 1.The sphere challenge: activity recognition with multimodal sensor data. http://blog.drivendata.org/2016/06/06/sphere-benchmark/
- 2.Agarwal, A., Rakhlin, A., Bartlett, P.: Matrix regularization techniques for online multitask learning. Technical report UCB/EECS-2008-138, EECS Department, University of California, Berkeley (2008)Google Scholar
- 5.Carós, J.S., Chételat, O., Celka, P., Dasen, S., CmÃral, J.: Very low complexity algorithm for ambulatory activity classification. In: Proceedings of the 3rd European Medical and Biological Conference, EMBEC, pp. 16–20 (2005)Google Scholar
- 8.Gao, Q., Doering, M., Yang, S., Chai, J.: Physical causality of action verbs in grounded language understanding. In: Proceedings of the 54th Annual Meeting of the Association for Computational Linguistics, vol. 1, pp. 1814–1824 (2016)Google Scholar
- 10.Lafferty, J., McCallum, A., Pereira, F.C.: Conditional random fields: probabilistic models for segmenting and labeling sequence data (2001)Google Scholar
- 12.Liu, X., Liu, L., Simske, S.J., Liu, J.: Human daily activity recognition for healthcare using wearable and visual sensing data. In: 2016 IEEE International Conference on Healthcare Informatics (ICHI), pp. 24–31. IEEE (2016)Google Scholar
- 15.Sutton, C., McCallum, A.: An introduction to conditional random fields for relational learning. In: Sutton, C., McCallum, A. (eds.) Introduction to Statistical Relational Learning, pp. 93–128. MIT Press, Cambridge (2006)Google Scholar
- 16.Twomey, N., Diethe, T., Kull, M., Song, H., Camplani, M., Hannuna, S., Fafoutis, X., Zhu, N., Woznowski, P., Flach, P., Craddock, I.: The SPHERE challenge: Activity recognition with multimodal sensor data. arXiv:1603.00797 (2016)
- 17.Vail, D.L., Veloso, M.M., Lafferty, J.D.: Conditional random fields for activity recognition. In: Proceedings of the 6th International Joint Conference on Autonomous Agents and Multiagent Systems, p. 235. ACM (2007)Google Scholar
- 18.Wang, S., Yang, J., Chen, N., Chen, X., Zhang, Q.: Human activity recognition with user-free accelerometers in the sensor networks. In: Proceedings of the International Conference on Neural Networks and Brain, vol. 2, pp. 1212–1217. IEEE (2005)Google Scholar
- 20.Yang, H., King, I., Lyu, M.R.: Online learning for multi-task feature selection. In: Proceedings of the 19th ACM International Conference on Information and Knowledge Management, pp. 1693–1696. ACM (2010)Google Scholar
- 21.Yang, S., Gao, Q., Liu, C., Xiong, C., Zhu, S.C., Chai, J.Y.: Grounded semantic role labeling. In: Proceedings of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies, pp. 149–159 (2016)Google Scholar