Daniel Anda Li
MIT Tang Family FinTech Undergraduate Research and Innovation Scholar
Large-Scale Dynamic Program Analysis Using Machine Learning
2023-2024
Electrical Engineering and Computer Science
- Artificial Intelligence & Machine Learning
Armando Solar Lezama
This project focuses on predicting part of the behavior of a computer program without having to run the program. Most of the previous works in this area involve analyzing code statically by using a model trained on a dataset of code without a lot of runtime information, while this project aims to perform dynamic analysis on large-scale real-world programs on the web. I will use a dataset that involves dynamic runtime data of JavaScript programs used on real websites. I will make a machine learning model trained on the dataset to perform the task. The model will take code as input and generate an output used for a query and an output used as context for other lines of code. The query part of the model will take as input the previous output and a query and output a yes/no answer.
I am participating in SuperUROP to get research experience in computer science and machine learning. I have taken machine learning courses and worked on a UROP involving machine learning. I hope to publish a paper and get additional research experience by the end of this SuperUROP.