Research and Publications | Cara's website

Research

First author of CLE-SMOTE: Addressing Extreme Imbalanced Data Classification with Contrastive Learning-Enhanced SMOTE
- Developed machine learning predictive models for classification datasets with severe class imbalance using different resampling techniques.
- Developed novel method to improve results from models trained with Synthetic Minority Oversampling Technique (SMOTE).
- Accepted and published at International Conference of Learning Representations (ICLR) DMLR workshop
First author of AQUASENT-TMMAE: a Self-supervised Learning Method for Water Quality Monitoring.
- Created state-of-the-art model using novel self-supervised learning method to detect acid mine drainage from satellite imagery.
- Published in KDD DeepSpatial Workshop (Summer 2024).
- One of 20 spotlight high school projects for NeurIPS (Fall 2024).

Schematic of AquaSent-TMMAE from published paper

Cluster: Mathematical Modeling of Biological Systems

Intensive four-week residential summer research program at UC Davis
Researched blended mathematical topics (differential equations, probability, evolutionary game theory) and biological topics (ecology, epidemiology, cell biology).
For final project, researched Ant Colony Optimization algorithm and created animated maze-solving simulation.

Global tech education non-profit organization that empowers girls to become tech entrepreneurs and leaders through coding mobile apps that address real-world problems
Programmer for CyberMentor, a mobile app that teaches digital literacy skills to young kids through games. App named Semifinalist in Technovation Girls app building challenge

Using satellite images taken after Hurricane Harvey, created program to help identify if a given region is likely to contain flooding damage