Heard on Campus: Combining machine learning and satellite images

UNIVERSITY PARK, Pa. — On Saturday, Feb. 4, Tamma Carleton, assistant professor of economics at the University of California, Santa Barbara, presented the third lecture in the 29th Ashtekar Frontiers of Science lecture series. This year’s lecture series focuses on how researchers are using and sharing “big data” to address longstanding scientific questions and make important societal contributions. The series is titled “Exploring Open Science and Big Data.”

In her talk, titled “Combining Satellite Imagery with Machine Learning to Address Global Challenges,” Carleton discussed how the combination of satellite imagery and machine learning has begun to transform our ability to map, monitor, and influence many global challenges, ranging from deforestation to poverty eradication to illicit activity. This emerging research area is data intensive and computationally demanding, making participation difficult for many researchers, governments, and nongovernmental organizations. 

“The idea is that with machine learning we are basically training computers to do the same thing that our human eye is doing,” Carleton said. “There potentially are things that computers can see in the images that we are not going to be able to predict.”

Carleton is therefore developing tools and methods that use machine learning to extract information from satellite images to help fill the large data gaps that exist for basic economic, social, and environmental indicators, which are highly unequally distributed across the world.

“At home in the United States, we have the American community survey, and we are surveying our populations  at least every year,” said Carleton. “Key regions of the world where some of our economic challenges are the largest are places where we're not seeing surveys for over 25 years.”

According to Carleton, in areas of the world where these surveys are scarce, we may be able to use satellite images and translate the imagery to information that can help us solve social, environmental and economic challenges. There are over 700 earth observation satellites that produce over 100 TB of data every single day.

“This massive amount of information that we're collecting from space can help us fill these data gaps and sort of level the playing field by providing a globally comprehensive — at least somewhat and we can talk about a bias — and in some ways objective picture of what's happening across the entire world.”

She then focused on new algorithmic innovations that make this field more accessible to a wider array of users with the aim to democratize access to this powerful new source of global information.

Carleton’s research combines economics with data sets and methodologies from remote sensing, data science and climate science to quantify how environmental change and economic development shape one another. Her current work focuses on climate change, water scarcity, and the use of remote sensing for global-scale environmental and socioeconomic monitoring. Carleton is a member of the Climate Impact Lab, a research associate at the Environmental Markets Lab, and a faculty research fellow at the National Bureau of Economic Research. She holds a doctoral degree in agricultural and resource economics from the University of California, Berkeley, and master’s degrees in environmental change and management as well as economics for development from the University of Oxford.

The final two lectures in the series will occur on Saturday, Feb. 11, and Saturday, Feb. 18, in 100 Thomas Building on the Penn State University Park campus. 

Feb. 11

Bing Pan, professor of commercial recreation and tourism in the College of Health and Human Development at Penn State, will present a lecture titled “Big Data Analytics in Tourism and National Park Research.” 

Pan will discuss how visitors to a destination or a national park interact with information technologies throughout their journeys and leave various online digital traces. He will then explain how search engine queries, website logs, mobile phone data, reservation data, social media, GPS traces, and simulations can help us monitor, predict and manage visitors to a place or a national park, understand their demographics and visitation experience, or find the best evaluation strategy in a national park in an emergency scenario. 

Feb. 18

Aleksandra Slavkovic, associate dean for graduate education in the Eberly College of Science and professor of statistics and public health sciences at Penn State, will present a lecture titled “The Role of Statistical Data Privacy in Support of Open Science and Public Policy.” 

In her lecture, Slavkovic will discuss the vast amount of sensitive data (e.g., health, financial, genomic, survey data) that is collected and archived by corporations, government agencies, health networks, and social networking websites. The social benefits of analyzing these data are significant and include support for open data access and reproducibility. However, the release of these data and/or analyses can be devastating to the privacy of individuals and organizations. Slavkovic will give an overview of challenges associated with protecting confidential data and will also discuss how integrating tools from statistics and computer science can provide formal privacy protection and thus address some of these challenges. 

About the Ashtekar Frontiers of Science Lectures

The Penn State Lectures on the Frontiers of Science was founded by Abhay Ashtekar in 1995, soon after he arrived at Penn State as director of a new research center that subsequently evolved to become the Institute for Gravitation and the Cosmos. It owes its success to tireless efforts and meticulous planning by Barbara Kennedy, who presided over the series during its first 25 years, making it one of the most successful science outreach events in central Pennsylvania.