Jasmine Zhang ’24
Using Machine Learning to Analyze Earth’s First Reefs
I investigated the fossils of 500 million-year-old reef-building organisms called archaeocyathids. To better understand the role that modern coral reefs play in mitigating climate change, we need to examine their prehistoric counterparts to determine the structure of these organisms and their interactions with their environment. My goal was to create an automated machine learning system known as a convolutional neural network to identify pixels of archaeocyathids from images of fossil samples. Before training the neural network, I created training data by segmenting and labeling archaeocyathid cross-section images. After training the network, I adjusted different parameters to improve its predictions with the goal of creating a 3D model of the archaeocyathids. As a result of this internship, my knowledge of computer vision and machine learning greatly increased, and I now have significant experience using MATLAB and Python software. I’m incredibly grateful to this internship for providing me with an informative and thought-provoking introduction to both geoscience and research, two areas I am excited to continue exploring.
Climate and Environmental Science
Maloof Research Group, Department of Geosciences, Princeton University
Adam Maloof, Professor of Geosciences; Ryan Manzuk, Ph.D. candidate, Geosciences