My research focuses on improving deep time series methods and applying them to challenging problems in health and biology.
Previously, I completed my BSc and MSc at UofT in CS and worked as an SDE at Amazon. This summer, I interned at Pinterest Labs working on NER.
PAN-cODE: COVID-19 forecasting using Conditional Latent ODEs
We apply the Latent ODE architecture for the task of short-term COVID-19 caseload forecasting. We condition latent trajectories on government interventions to allow alternative caseload trajectory generation. Our method is out performs or is comparable to the state-of-the-art models when applied to US regions.
Journal of the American Medical Informatics Association, 2022
Segmenting Hybrid Trajectories using Latent ODEs
We developed LatSegODE to represent hybrid trajectories (those with discontinuous jumps and dynamical mode changes) using Latent ODEs. After fitting Latent ODEs on trajectory primitives, we use changepoint detection methods to find optimal locations to restart ODE dynamics. Reconstructions are scored using the marginal likelihood to regularize the number of detected changepoints.
International Conference of Machine Learning (ICML) 2021
Reconstructing evolutionary trajectories of mutation signature activities in cancer using TrackSig
We developed a method to reconstruct mutational signature trajectories in cancer populations through time. Mutational activity was determined using a mixture of multinomials, and joint optimal segmentation was used to segment trajectories into distinct mutagenic periods.
Nature Communications, 2020
ePlant : Visualizing and Exploring Multiple Levels of Data for Hypothesis Generation in Plant Biology
We developed an analytic portal for plant model species, integrating over 12 different levels of visualizations on millions of data points. I developed the Protein Interactions Viewer, displaying known and predicted interactions between proteins and DNA.
Plant Cell, 2017
Predicting Patient Reported COPD Symptoms Using Smart Device Sensor Data
We applied various classical and deep learning methods to predict the onset of symptoms of COPD exacerbation from smartwatch sensor data. Developed as part of CSC2541, with data obtained from the WearCOPD project.
Clustering Subclonal Phylogenies Using Gaussian Mixture Models
We applied clustering methods on phylogenetic tree reconstructions of cancer tumors. Non-parametric GMM clustering was applied on MCMC sample outputs to obtain characteristic phylogentic tree reconstructions from PhyloWGS.
- CSC311: Intro to Machine Learning (2020-Present)
- CSC413: Neural Networks and Deep Learning (2020-2022)
- CSC207: Software Design (2019)