BY MEG LOPES
Karianne Bergen, Assistant Professor of Data Science and Earth, Environmental, and Planetary Sciences at Brown recently sat down with Sam Charrington on his podcast, "The TWIML AI Podcast" to discuss machine learning for earthquake seismology.
She discusses her journey as a researcher, from her undergraduate years at Brown to working in a lab and ultimately getting her Ph.D. Karianne's work joins machine learning approaches and earthquake seismology. Now, she's back at Brown and teaching data science skills to future researchers.
Karianne completed her undergraduate degree through the Division of Applied Mathematics at Brown. After spending some time working for MIT Lincoln Laboratory where she worked on multi-sensor data fusion, Karianne went to Stanford University for her Ph.D.
“I went to Stanford and everyone's talking about data science. If you're doing computational and applied mathematics, the natural place to apply that is in data science and machine learning. I ended up moving more in that direction than I had anticipated."
After discovering a passion for data science, Karianne worked with a Stanford researcher on a seismology project that ended up being her dissertation. In her interview, she talks about some of the problems that arise when applying machine learning techniques to seismic events.
"In a lot of earth science applications, in sciences in general, it's common to have really big data sets but you don't always have labels for them or a very good set of labels. And especially if your goal is to find something new, then the labeled data you have may not be so helpful for finding and discovering new things in the data. So we wanted to develop something that would be able to discover new earthquake signals in these continuous data sets without having to have a lot of labeled data. This continues to be a need in this field because labeled data is something that people often don't have."
After finishing her dissertation, Karianne wondered how scientists could better incorporate machine learning in the field of earthquake seismology. She published a paper in 2018 that included recommendations for the scientific community on how data scientists can use machine learning models to better detect earthquakes and interpret data sets in Earth geoscience.
Karianne returned to Brown in 2021 as a new faculty member. She continues to collaborate with data scientists at Brown, researching seismograms, signal processing, and sensor networks through her interdisciplinary appointments. With the growing interest in the field of scientific machine learning, Karianne hopes to develop new ideas and new tools through her research, while training Brown students interested in the intersection of seismological data and machine learning.
"My opinion is that Brown undergraduates are the most fun and greatest undergraduates in the world, so I was excited to come back and teach them."
Watch Karianne's full interview below.