Projects

Research

Multivarite Time Series Classification with Random Forest: Currently extending random forests with structured projection matrices for applications in EEG signal classification problems related to decision making and anatomical classification. (Results to come!)
Semi-Automatic Localization of Stereo-EEG Contacts: Developed an algorithm to segment contacts from CT images of epilepsy patients to streamline the coregistration process. (View project on Github)

Music Translation (View project on Github): A team project that combined the WaveNet autoencoder architecture with Contrastive Predictive Coding (CPC) encoding to translates audio snippets of classical music from one instrument to another among three selected instruments: piano, violin, and cello. See a brief manuscript here and an article about it here. This project was done in collaboration with @Maggie Wang, @Silu Men, and @David Shi.
Tricuviz: A team project that performs near real-time registration of pre-operative CT scans with intraoperative fluoroscopy images for transcatheter tricuspid valve replacement operations. Code is private.
Matrix Analysis Notes (View project on Github): Transcribed course notes for AMS792 Matrix Analysis (Fall 2020) into TeX.
Tetris AI (View project on Github): Developed a Deep Q-Network using the deep reinforcement learning framework implemented using Python and TensorFlow.
Tri-Tag (View project on Devpost): A modern, digital approach to Triage. My group received a Top Ten in Category award out of hundreds of participandts at MedHacks 2018. TriTag is an efficient patient marker and tracking system for patients in Emergency and Mass Casualty events. It consists of three steps: (1) Initial assessment and assignment of severity and hash code, (2) HIPAA-compliant cloud database of important medical data for remote retrieval, and (3) ambulatory tracking of patients through cloud database.