Arthur Martins de Souza Barreto

Expected Graduation Date
2025

Major
Computer Engineering

Research Interests
Artificial Intelligence Interactive Computing Scientific Computing Security and Privacy

Research Mentor
Lawrence Rauchwerger

Research Description
Parallel Box Least Square Algorithm for Exoplanets Detection

Biography

I am involved in two scientific initiation programs at Insper, one locally with my advisor, Luciano Silva, and another in partnership with the University of Illinois. Both programs are focused on the field of Astroinformatics and aim to improve software in this area.

My first research project involves the use of quantum computing to enhance neural network training. Specifically, I am working on the identification of exoplanets, which currently takes a considerable amount of time to determine whether there is a transit in a light curve file. To tackle this issue, I have spent significant time understanding the library Lightkurve, and then delved into the Qiskit library to learn about quantum computing. These steps took me almost a year to complete, and I have recently developed my own code to determine the presence of a transit in a light curve file. Furthermore, I have evaluated the code and derived an equation to correct its results in relation to the Lightkurve library.

Additionally, I have compiled a dataset consisting of 200 star positions from the Kepler satellite and labeled them as transit or non-transit. Currently, I am training a classical neural network to determine if there is a transit in a given light curve file, which typically contains approximately 50,000 data points. The next step in this project is to identify where the neural network training is taking too long, and then apply quantum computing to expedite this process. Ultimately, my goal is to have a neural network trained using quantum computing.

My second project, in partnership with Professor Lawrence and student Selin Yildirim at the University of Illinois, is focused on understanding why the C library used by NASA to find exoplanets takes too long to perform. Professor Lawrence has created a library, STAPL, that verifies the parts of the code that take too long to run. Since the beginning of this year, we have been focusing on identifying those parts that take too long to run and, in a future iteration, improving them.

In conclusion, I would like to research topics related to my previous research in the training of neural networks. Additionally, due to my familiarity with robotics, I would be interested in exploring projects that combine both fields, or even using neural networks to find potential treatments and cures for diseases.