Projects will typically be under mentoring of a graduate student. Some of the project options within our group are listed below.
- Core ML research: In this option, you will join an ongoing research project. The goal is generating original research resulting in a high-quality publication. Undergraduate students are expected to contribute to implementation and algorithms.
- ML for Hydrology (joint with Dr. Hoori Ajami): Hydrological models aid in understanding, predicting, and managing water resources. To accelerate modeling we would like to design learning-based surrogate models to approximate the behavior of the physics-based simulations and offer a more viable solution for efficient data integration and prediction compared to ParFlow.
- Cryptocurrency price manipulation: Cryptocurrency pump-&-dump schemes are hurting many unsuspecting Americans. These schemes often manipulate social media to “hype up” their project. How can we detect them on social media or price data? Can we setup Twitter bots that automatically inform people before such schemes.
- Understanding Reddit: Reddit has a huge variety of discussion groups (aka subreddit). Can we train NLP models (such as BERT) and infer interesting things about Redditors? Of specific interest is finance subreddits such as Wallstreetbets (which relates to the crypto topic) as well as understanding what makes certain posts receive a lot of upvotes.