Real-time clustering of mouse paw trajectory, with trajectory obtained from DeepLabCut.
Trajectory clustering is primarily done using TRACLUS (Lee, Han, & Whang, 2007), with adaptations to work in a real-time setting.
From this clustering, the mouse will be rewarded for specific paw movement, allowing for automated training of paw motion.
Lee, J., Han, J., & Whang, K. (2007). Trajectory clustering. Proceedings of the 2007 ACM SIGMOD International Conference on Management of Data - SIGMOD 07. doi:10.1145/1247480.1247546