Abstract
This article presents an in-depth experimental study of Ensemble Deep Learning techniques on small datasets for the analysis of time-series data generated by wearable devices. Deep Learning networks generally require large datasets for training. In some health care applications, such as the real-time smartwatch-based fall detection, there are no publicly available, large, annotated datasets that can be used for training, due to the nature of the problem (i.e., a fall is not a common event). We conducted a series of offline experiments using two different datasets of simulated falls for training various ensemble models. Our offline experimental results show that an ensemble of Recurrent Neural Network (RNN) models, combined by the stacking ensemble technique, outperforms a single RNN model trained on the same data samples. Nonetheless, fall detection models trained on simulated falls and activities of daily living performed by test subjects in a controlled environment, suffer from low precision due to high false-positive rates. In this work, through a set of real-world experiments, we demonstrate that the low precision can be mitigated via the collection of false-positive feedback by the end-users. The final Ensemble RNN model, after re-training with real-world user archived data and feedback, achieved a significantly higher precision without reducing much of the recall in a real-world setting.
- Fadel Adib, Hongzi Mao, Zachary Kabelac, Dina Katabi, and Robert C. Miller. 2015. Smart homes that monitor breathing and heart rate. In Proceedings of the 33rd Annual ACM Conference on Human Factors in Computing Systems (CHI’15). ACM, New York, NY, 837--846. DOI:https://doi.org/10.1145/2702123.2702200Google Scholar
- Apple Inc. 2019. Apple Watch Series 4. Retrieved from http://www.apple.com/apple-watch-series-4/activity/.Google Scholar
- Joseph D. Bronzino. 2006. Biomedical signals: Origin and dynamic characteristics; frequency-domain analysis. In Medical Devices and Systems. CRC Press, 27--48.Google Scholar
- Eduardo Casilari, Raul Lora-Rivera, and Francisco Garcia-Lagos. 2020. A study on the application of convolutional neural networks to fall detection evaluated with multiple public datasets. Sensors 20, 5 (2020). DOI:https://doi.org/10.3390/s20051466Google Scholar
- S. Chaudhuri, H. Thompson, and G. Demiris. 2014. Fall detection devices and their use with older adults: A systematic review. J. Geriatric Phys. Ther. 37, 4 (2014), 178--196.Google Scholar
Cross Ref
- Xuhui Chen, Jinlong Ji, Tianxi Ji, and Pan Li. 2018. Cost-sensitive deep active learning for epileptic seizure detection. In Proceedings of the ACM International Conference on Bioinformatics, Computational Biology, and Health Informatics (BCB’18). ACM, New York, NY, 226--235. DOI:https://doi.org/10.1145/3233547.3233566Google Scholar
- Adele Cutler, D. Richard Cutler, and John R. Stevens. 2012. Random Forests. Springer, Boston, MA, 157--175. DOI:https://doi.org/10.1007/978-1-4419-9326-7_5Google Scholar
- Jerome H. Friedman. 2002. Stochastic gradient boosting. Comput. Stat. Data Anal. 38, 4 (2002), 367--378.Google Scholar
Digital Library
- Stephen Grossberg. 2013. Recurrent neural networks. Scholarpedia 8, 2 (2013), 1888.Google Scholar
Cross Ref
- Yu Guan and Thomas Plötz. 2017. Ensembles of deep LSTM learners for activity recognition using wearables. Proc. ACM Interact. Mob. Wearable Ubiquitous Technol. 1, 2, Article 11 (June 2017), 28 pages. DOI:https://doi.org/10.1145/3090076Google Scholar
Digital Library
- Yu Guan and Thomas Plötz. 2017. Ensembles of deep LSTM learners for activity recognition using wearables. Proc. ACM Interact. Mob. Wearable Ubiquitous Technol. 1, 2, Article 11 (June 2017), 28 pages. DOI:https://doi.org/10.1145/3090076Google Scholar
Digital Library
- Ngu Anne Hee, Wu Yeahuay, Zare Habil, Polican Andrew, Yarbrough Brock, and Yao Lina. 2017. Fall detection using smartwatchsensor data with accessor architecture. In Proceedings of the International Conference for Smart Health (ICSH’17).Google Scholar
- Andreas Janecek, Wilfried Gansterer, Michael Demel, and Gerhard Ecker. 2008. On the relationship between feature selection and classification accuracy. In New Challenges for Feature Selection in Data Mining and Knowledge Discovery. 90--105.Google Scholar
- Diederik P. Kingma and Jimmy Ba. 2014. Adam: A method for stochastic optimization. Retrieved from https://Arxiv:1412.6980.Google Scholar
- Alex Krizhevsky, Ilya Sutskever, and Geoffrey E. Hinton. 2012. Imagenet classification with deep convolutional neural networks. In Advances in Neural Information Processing Systems. MIT Press, 1097--1105.Google Scholar
Digital Library
- Bogdan Kwolek and Michal Kepski. 2014. Human fall detection on embedded platform using depth maps and wireless accelerometer. Comput. Methods Programs Biomed. 117, 3 (2014), 489--501.Google Scholar
Digital Library
- Song-Mi Lee, Sang Min Yoon, and Heeryon Cho. 2017. Human activity recognition from accelerometer data using convolutional neural network. In Proceedings of the IEEE International Conference on Big Data and Smart Computing (BigComp’17). IEEE, 131--134.Google Scholar
- Taylor R. Mauldin, Marc E. Canby, Vangelis Metsis, Anne H. H. Ngu, and Coralys Cubero Rivera. 2018. SmartFall: A smartwatch-based fall detection system using deep learning. Sensors 18, 10 (2018). DOI:https://doi.org/10.3390/s18103363Google Scholar
- Daniela Micucci, Marco Mobilio, and Paolo Napoletano. 2017. UniMiB SHAR: A dataset for human activity recognition using acceleration data from smartphones. Appl. Sci. 7, 10 (2017). DOI:https://doi.org/10.3390/app7101101Google Scholar
- Mirto Musci, Daniele De Martini, Nicola Blago, Tullio Facchinetti, and Marco Piastra. 2018. Online fall detection using recurrent neural networks. Retrieved from https://Arxiv:1804.04976.Google Scholar
- Henry Friday Nweke, Ying Wah Teh, Mohammed Ali Al-garadi, and Uzoma Rita Alo. 2018. Deep learning algorithms for human activity recognition using mobile and wearable sensor networks: State of the art and research challenges. Expert Syst. Appl. 105 (2018), 233--261. DOI:https://doi.org/10.1016/j.eswa.2018.03.056Google Scholar
Cross Ref
- RightMinder. 2018. RightMinder—Fall Detection for Android Smartwatches and Android Phones. Retrieved from http://www.rightminder.com.Google Scholar
- H. M. Salman, W. F. Wan Ahmad, and S. Sulaiman. 2018. Usability evaluation of the smartphone user interface in supporting elderly users from experts? perspective. IEEE Access 6 (2018), 22578--22591. DOI:https://doi.org/10.1109/ACCESS.2018.2827358Google Scholar
Cross Ref
- Guto Leoni Santos, Patricia Takako Endo, Kayo Henrique de Carvalho Monteiro, Elisson da Silva Rocha, Ivanovitch Silva, and Theo Lynn. 2019. Accelerometer-based human fall detection using convolutional neural networks. Sensors 19, 7 (2019). DOI:https://doi.org/10.3390/s19071644Google Scholar
- Jose Antonio Santoyo-Ramon, Eduardo Casilari, and Jose Manuel Cano-Garcia. 2018. Analysis of a smartphone-based architecture with multiple mobility sensors for fall detection with supervised learning. Sensors 18, 4 (2018). DOI:https://doi.org/10.3390/s18041155Google Scholar
- S. Shin and W. Sung. 2016. Dynamic hand gesture recognition for wearable devices with low complexity recurrent neural networks. In Proceedings of the IEEE International Symposium on Circuits and Systems (ISCAS’16). 2274--2277. DOI:https://doi.org/10.1109/ISCAS.2016.7539037Google Scholar
- Angela Sucerquia, José David López, and Jesús Francisco Vargas-Bonilla. 2017. SisFall: A fall and movement dataset. Sensors 17, 1 (2017), 198.Google Scholar
Cross Ref
- C. Tacconi, S. Mellone, and L. Chiari. 2011. Smartphone-based applications for investigating falls and mobility. In Proceedings of the 5th International Conference on Pervasive Computing Technologies for Healthcare (PervasiveHealth) and Workshops. 258--261. DOI:https://doi.org/10.4108/icst.pervasivehealth.2011.246060Google Scholar
- T. Theodoridis, V. Solachidis, N. Vretos, and P. Daras. 2018. Human fall detection from acceleration measurements using a recurrent neural network. In Precision Medicine Powered by pHealth and Connected Health. Springer, 145--149.Google Scholar
- Jindong Wang, Yiqiang Chen, Shuji Hao, Xiaohui Peng, and Lisha Hu. 2019. Deep learning for sensor-based activity recognition: A survey. Pattern Recogn. Lett. 119 (2019), 3--11. DOI:https://doi.org/10.1016/j.patrec.2018.02.010 Deep Learning for Pattern Recognition.Google Scholar
Digital Library
- S. Yu, L. Qin, and Q. Yin. 2018. A C-LSTM neural network for human activity recognition using wearables. In Proceedings of the International Symposium in Sensing and Instrumentation in IoT Era (ISSI’18). 1--6. DOI:https://doi.org/10.1109/ISSI.2018.8538129Google Scholar
- Cha Zhang and Yunqian Ma. 2012. Ensemble Machine Learning: Methods and Applications. Springer.Google Scholar
- Wang Zhiqiang and Liu Jun. 2017. A review of object detection based on convolutional neural network. In 2017 36th Chinese Control Conference (CCC'17). IEEE, 11104--11109.Google Scholar
- Ahmet Turan Ozdemir. 2016. An analysis on sensor locations of the human body for wearable fall detection devices: Principles and practice. Sensors 16, 8 (2016). DOI:https://doi.org/10.3390/s16081161Google Scholar
Index Terms
Ensemble Deep Learning on Wearables Using Small Datasets
Recommendations
Detecting unseen falls from wearable devices using channel-wise ensemble of autoencoders
Investigated the use of Autoencoders to learn generic features from wearable devices.Proposed channel-wise ensemble approaches for Autoencoders to identify unseen falls.Developed new threshold tightening approaches on reconstruction error of ...
Elderly Assistance Using Wearable Sensors by Detecting Fall and Recognizing Fall Patterns
UbiComp '18: Proceedings of the 2018 ACM International Joint Conference and 2018 International Symposium on Pervasive and Ubiquitous Computing and Wearable ComputersFalling is a serious threat to the elderly people. One severe fall can cause hazardous problems like bone fracture or may lead to some permanent disability or even death. Thus, it has become the need of the hour to continuously monitor the activities of ...
A Survey of Fall Detection Algorithm for Elderly Health Monitoring
BDCLOUD '15: Proceedings of the 2015 IEEE Fifth International Conference on Big Data and Cloud ComputingWith the arrival of population aging society, the health care of the elderly becomes more important. The fall detection algorithm is the core of the fall detection alarm system, so it is the key for the research and development of the fall detection ...






Comments