Projects

“Us” aims to help people develop their sense of empathy. It uses machine learning to analyze interlocutors’ signals (e.g., tone of voice, choice of words, facial expressions, electrodermal activity, heart rate, etc.) during conversations, and feeds the emotional content back to them in real time and through multiple analyses. We have versions of the system for both in-person and online conversations.

Our contribution is that we brought OpenMP implementation to the kernel in three different approaches - runtime in kernel (RTK), process in kernel (PIK) and custom compilation for kernel (CCK). Notably, the average performance gain of our RTK approach across scales and benchmarks is on the order of 22% (geometric mean) compared to Linux user-level code.

We aimed to invent novel methods to help scientists easily locate and identify 3-D microscopic biological structures from large amount of unlabeled Cryo-Electron Tomography data, using Computer Vision techniques.

We have built a complete robot control system without big, heavy wearables. We control our robot by standing in front of MS Kinect v2. We also integrated VR and object detection technologies to make our robot system intelligent, so the robot operator will have an immersive experience while controlling and playing with our system.