Anubhav Agarwal ’23

Major

Computer Science

Project Title

Application of New Machine Learning Tools for the Study of Ooids and Their Ancient Tropical Marine Environments

Certificate(s): Statistics and Machine Learning, Applied and Computational Mathematics

I worked with ooids, which are small sedimentary grains that form in shallow bodies of water. Ooids go through several size changes due to many factors such as currents or ocean chemical composition. My work focused on automating the analysis of 2D images of ancient giant ooids captured from the rock record. I worked on using a fully convolutional neural network to separate each individual ooid from the background in the image. Most of my summer was spent working to optimize the region proposal network and the inside/outside classifiers aspects of this network for the ooid problem. Throughout this process, I learned the advancements that have been made from the most basic convolutional networks, to the cutting-edge work that’s done in computer vision today. I also realized the increased difficulty in optimizing networks of this complexity. Overall, this internship was an amazing opportunity that cemented my desire to work with complex machine networks as I pursue independent research in the future.



Internship Year

2021

Project Category

Climate and Environmental Science

Organization(s)

Maloof Research Group, Department of Geosciences, Princeton University

Mentor(s)

Adam Maloof, Professor of Geosciences; Bolton Howes, Ph.D. candidate, Geosciences