Projects in Grad School

In my PhD, I've done some fun projects for courses as well as for my own research. Here's a brief summary of some of them:

I reproduced the numerical results from "Cautious Model Predictive Control using Gaussian Process Regression " Using a kinematic bicycle model with Pacejka tire model for friction, the goal is to learn additive dynamics that are not initially known. The learning is performed using Gaussian process regression, which not only produces an expected value, but also uncertainty quantification in the form of a variance estimate. The predictive control uses chance-constraints to act on the uncertainty quantification. The resulting controller solves on the order of milliseconds but decreases in speed as the size of the training dataset increases, a known shortcoming of this regressor. These results have motivated further works of my own that are still in the pipeline.

Two cars race (in simulation) against each other, each with the goal of winning. We explored the zero-sum setting, and used iterated best response in order to actually compute solutions for this game.

Fairness in machine learning is an important and exciting area of research. My teammate and I looked at the COMPAS dataset on predicting inmate recidivism, and built a classifier that would "more fairly" predict recidivism rates.

More to come when I have more time to share...