By: Agnes Villaroman

Agnes Villaroman: What inspired you to do this project? 

Silas Gifford: I was able to go out with Sarah Codde. She goes out more than seventy times a year to monitor elephant seals and count them. So I got to go with her one day and it was a full morning of walking up and down beaches, scrambling over walks trying not to get my shoes wet. It was cool to be that close to elephant seals—they were big. I had this special experience of being ten feet away from the seals and I really, really enjoyed that experience and understood how time, resource, and people intensive it is to monitor these elephant seals. So Ben asked, “Can you apply machine learning to this problem?” and I said “Yeah!”    

Villaroman: What are some things you learned from your research? Anything surprising?

Gifford: Everybody thought elephant seals were extinct, until Smithsonian explorers found eight elephant seals after years of people thinking they were gone forever. However, what the exhibition did was, the explorers killed seven of the eight elephant seals to bring back to the museum. So I think there’s a lot of themes of both death and rebirth because of this history of seal hunting to extinction, and even when we found they weren’t extinct we did not do our duty to protect them.  

Villaroman: How would you bridge the gap from your research to research users (science communication)?

Gifford: There are safe instances where you can look at elephant seals. Some beaches allow you to be close enough where people can see them and smell them—these are stinky animals! You can see how they interact with each other. The awesome thing about deep learning today is, for the most part, we have frameworks where you can apply pretty minimal prior knowledge so you can figure out how our machine learning model works. We can share our models and data sets very well. 

Villaroman: What is some advice you have for students pursuing a project in a similar field as you?

Gifford: The Data Scholars and Data Discovery Programs are very useful. The most important thing is to apply to different projects and talk to different people and ask, “I see you are doing X thing and I’m really interested, could you use one more person?” All my most important and useful pathways into projects were because of my network. Not everyone is going to say yes, but there are enough people who need data scientists that someone is bound to say “Yes.”