Graduate Student Spotlight: Daniel Cheng’s Computer Science Research Has Real-World Impact

When Daniel Cheng first came to UCI in 2013, he wanted to pursue a bachelor’s degree in computer science because of its “applicability to many other fields.” Now, as a Ph.D. student in the Donald Bren School of Information and Computer Sciences (ICS), he’s practicing that applicability with real-world impact. Specifically, he has led the development of the Calving Front Machine (CALFIN), a program that uses artificial intelligence, machine learning and neural networks to help Earth science researchers monitor glacier loss. He collaborated on CALFIN with researchers from UCI, the University of Washington, and NASA’s Jet Propulsion Laboratory (where he’s an intern), and the team outlined their work in a paper published by The Cryosphere. Their efforts were further spotlighted in “An Artificial Neural Network Joins the Fight Against Receding Glaciers,” an article pointing out that this “neural network is capable of recognizing and measuring the edges of glaciers in satellite images of Earth’s surface nearly as well as a trained scientist, except it’s autonomous, quick, and can reliably process countless more glaciers than any human being ever could.” Here, Cheng talks about his interdisciplinary research and how computer science can help us better understand climate change.

What brought you to UCI with an interest in computer science?
I started at UC Irvine pursuing a B.S. in computer science, which initially interested me due to its applicability to many other fields. It gave me options to choose specialties even as I developed my skills in research during my bachelor’s and later master’s degrees here at UCI. I was able to get involved in undergraduate research due to the outreach efforts of my current adviser, Professor Wayne Hayes, which has developed into interdisciplinary studies in the Earth sciences as I pursue my Ph.D. Now, I get to apply CS concepts like artificial intelligence, machine learning (AI/ML) and neural networks (programs that “learn” to recognize patterns you teach it) to real-world problems, such as tracking changes in our climate over the last few decades.

What motivated you to develop CALFIN?
After getting introduced to other researchers, we were able to identify a need for automating a certain task that was both useful and time consuming. This task was the labeling of calving fronts, or where glaciers break off (“calve”) into the ocean. It’s important to track these changes in glaciers over the past decades, so that we can understand and even model how the Earth’s changing climate affects sheets of ice in places like Greenland and Antarctica. Automating this labeling task is perfectly suited to AI/ML, and neural networks in particular, which is what motivated me to develop CALFIN.

How are you continuing to improve and expand CALFIN?
I’m looking to improve CALFIN’s capabilities to detect additional glacier features, such as where water gathers in melt ponds on top of glaciers, or where cracks and rifts appear on the glacier surface. I’m also looking to expand CALFIN, as we’ve been focused on glaciers in Greenland but are already seeing promising results in extracting useful information from ice sheets in Antarctica.

How can computer science and these kinds of multidisciplinary projects help address climate change?
The intersection between computer science and other fields such as Earth science is large, and growing larger. We have a lot of data but a limited amount of people to process and analyze it. Computer science, through new techniques like AI/ML, will allow us to make use of our data to help understand and model climate change, which in turn will allow us to make decisions that affect our lives in the immediate and near future.

What are your future plans?
My future plans are to pursue a postdoctoral position to continue applying AI/ML to the Earth sciences, though my dissertation and defense will come first of course!

Are there plans to apply your technique for artificial neural networks to other application areas?
I’d like to continue developing these neural networks for Earth science applications, though beyond that scope, I can’t say for sure. I’m sure there’s a lot of room for growth and research that many others in the field can tackle, and it’ll be exciting to see what others come up with in the near future!

— Shani Murray