Projects

Accelerating Flow-Based Sampling for Large-N Gauge Theories

Research completed under the supervision of Fernando Romero-López and professor Phiala Shanahan. This was my thesis for my master's degree.

Deep Learning and Symbolic Regression for Discovering Parametric Equations

Research completed under the supervision of Samuel Kim and professor Marin Soljačić. Appeared in AI for Science (ICML 2022). [Link]

6.860/9.520 final project: Empirical evaluations of learning datasets with weight-sharing and locality.

This project provides an emperical evaluation of the difficulties of learning weight-sharing and locality in datasets satisfying both assumptions. [Link]

6.867 final project: ClassroomNet: Learning generalizable, interpretable features via knowledge distillation.

This project proposes and attempts to implement a neural network architecture in which 3D-to-2D knowledge distillation is used to learn better features in image processing tasks. [Link]

6.854/18.415 final project: Algorithms for fast selection on heap data structures.

This project surveys the theory behind the soft heap data structure and its uses in selection algorithms. In addition, the selection algorithms were implemented and their performances were compared. [Link]