My research broadly involves learning representations for time-series data, with a special focus on developing techniques that require minimal supervision. I develop unsupervised and self-supervised learning algorithms for time series data collected using sensors such as accelerometers and gyroscopes on wearables including smartwatches/smartphones.
I am currently working on learning unsupervised representations for performing human activity recognition using sensory data from wearables such as smartwatches.
The goal of my research is to ultimately reduce reliance on labeled data, due to the expensive and time-consuming nature of the annotation process. The idea is to make economic use of the available labeled data (however small) in the best way possible by leveraging larger quantities of unlabeled data.