Wearable Sensor Data Classification for Human Activity Recognition (HAR)
Key Personnel: Kandethody Ramachandran, Yicheng Tu, Ming Ji
Sensor based human activity recognition has applications in areas such as healthcare monitoring, sports, physical fitness, etc. Major purpose of such efforts is to achieve noninvasive and mobile activity monitoring. Using ubiquitous, cheap and widely available technology is the key requirement for human activity recognition. One of the projects we are looking at involves creation of a platform to combine off-the-shelf sensors of smartphones and smartwatches for recognizing human activities. We will develop novel computational algorithms so as to identify the activities in real time. In order to achieve the best tradeoff between the system’s computational complexity and recognition accuracy, we will develop multiplel evaluations criteria to determine which classification algorithm and features to be used.
- Some Publications
· Shuang Na, Kandethody M Ramachandran, Ming Ji, and Yicheng Tu, Real-time Activity Recognition using Smartphone Accelerometer, Submitted to 7th Annual International Conference on Computational Mathematics, Computational Geometry & Statistics (CMCGS 2018)
· Shuang Na, Kandethody M. Ramachandran, and Ming Ji, Online Bayesian Kernel Segmentation and a application, Preprint, 2017.
Possible project areas
- Energy forecasting
- model for tourism what (which variable) contributes most
Research on the impacts of tourism in the state of Florida
– with 87.3 million visitors in 2011, Florida is the top travel destination in the world. The tourism industry has an economic impact of $67 billion on Florida’s economy.
- Health related
Clinical, community-based or practice-based research in areas of tobacco and alcohol prevention, cancer, ageing, obesity, mental health, air and water quality etc.
- Homeless what causes homeless develop a model
- Environmental based
- Real time data and dynamical systems (relevance)
- Individualized medicine
- Madicaid loses $65 billion fraud, Create a fraud detection syste