Hello! I'm a Computer Science PhD student at Princeton University, where I'm fortunate to be advised by Ryan P. Adams. My research interests are broadly in understanding symmetry and structure in machine learning, and developing theoretically grounded ML methods for applications in computational physics and materials science. Some research directions I am excited about include:
- Scalable methods for incorporating symmetry constraints into ML models
- Ways to learn (approximate) symmetries in the data
- ML approaches for scaling density functional theory (DFT) calculations
I did my undergrad at MIT, where I worked on machine learning research with Sarah Cen and Devavrat Shah. I also previously worked on software for Uber ATG, MIT App Inventor, and Divvly.
You can contact me at cindyz at princeton dot edu.
Publications
-
A Single Architecture for Representing Invariance Under Any Space Group
Cindy Y. Zhang, Elif Ertekin, Peter Orbanz, Ryan P. Adams
We construct a single neural architecture that is capable of adapting its weights automatically to enforce invariance to any input space group.
Preprint.
[Paper] -
Matrix Estimation for Individual Fairness
Cindy Y. Zhang*, Sarah H. Cen*, Devavrat Shah
We present a framework for using singular value thresholding to pre-process sparse, noisy data that improves individual fairness without sacrificing performance.
International Conference on Machine Learning (ICML) 2023.
[Paper]
Projects
-
Kirigami Strata: Layers of Symmetry and Form
Isabel Moreira de Oliveira, Rafael Pastrana, Cindy Y. Zhang, Sigrid Adriaenssens, Ryan P. Adams, Emily Baker, Vittorio Paris — fabricated by Taramelliand Carpenteria Bonatese
This project is part of the European Cultural Centre's architecture exhibition Time Space Existence in Venice, Italy. The tiling pattern for this art piece is one of many possible patterns generated by a flow-based model with enforced symmetries and physical constraints. This pattern is then cut, deployed, and connected from a thin flat steel sheet to form a kirigami space frame with enhanced load-carrying capacity and stiffness.
[Project Page] [Blog Post]