Internship Reflection

Perspectives on Research In Science & Mathematics (PRISM)

During the summer of 2023, I worked with Dr. Julian Dymacek as a part of the PRISM program. Our project was to use non-negative matrix factorization to extract patterns from images of avian nocturnal flight calls (NFCs) to try to classify the species of bird. This experience helped me grow because it helped me decide if research and graduate school was an avenue I would want to pursue, and also gave me an opportunity to perform research in my field for the first time. I definitely learned a lot through this program, not only did I have to learn how to use new Python libraries (Numpy, Sklearn, etc.), but I learned a lot about the process that computer scientist researchers use. Most of our workdays were spent writing a lot of smaller programs that eventually fit together in the grander scheme of the project. We also spent a lot of time thinking up new approaches to take when our current solution was not working, or not making any progress. My favorite “tool” that I wrote was a program that used simulated annealing to explore a parameter space and see what group of parameters gave us the best accuracy when classifying. I have an interest in genetic algorithms, and ways that we can use concepts from biological or physical processes to design algorithms, so having the chance to implement this was really exciting. All in all, the project was a much greater success than I ever thought. When we started, I thought we would be lucky to correctly classify a call fifty percent of the time, but we actually wound up over 80 percent with a couple weeks left!