The following story was adapted from a news release recently published by Naval Surface Warfare Center, Crane Division. Images are courtesy of NSWC Crane.
Indiana University and Naval Surface Warfare Center, Crane Division, recently collaborated on research into innovative techniques designed to protect artificial intelligence and machine learning systems used by the U.S. Navy from adversarial threats.
More than 20 IU undergraduate and graduate students participated in the research partnership, which was established between IU and NSWC Crane in late 2019 and enabled through the Naval Engineering Education Consortium. The NEEC is a part of the Naval Sea Systems Command, which supports project-based collaborations with colleges and universities.
NSWC Crane supports electronic, engineering and ordnance needs of the U.S. Navy and other military customers. It is one of Indiana’s largest high-tech employers with over 3,800 employees, of which 2,500 are scientists, engineers and technicians. NSWC Crane is the principal tenant command at Naval Support Activity Crane, the third-largest naval installation in the world, which is located approximately 40 minutes southwest of IU’s Bloomington campus.
David Crandall, Luddy Professor of Computer Science and director of the Luddy Center for Artificial Intelligence at the IU Luddy School of Informatics, Computing and Engineering, said the project greatly benefited the participating IU students, who were able to test their knowledge and apply their research training in a rapidly growing field with major implications for national security.
“NEEC has been a wonderful opportunity for my students to both work on cutting-edge research, and to receive training that they wouldn’t normally get in the classroom,” said Crandall. “It was also an opportunity for us to push interdisciplinary boundaries between fields, because this was a collaboration between Professor Katy Borner’s expertise in visualization and my lab’s expertise in computer vision and machine learning. Together, our groups investigated new techniques for visualizing, debugging and improving the machine learning algorithms that are at the heart of modern artificial intelligence.”
Tyler Fitzsimmons, an engineer at NSWC Crane and technical point of contact for the NEEC project, said the three-year effort will help the U.S. Navy fleet better understand if an adversary is manipulating data and defend AI algorithms against malicious attempts.
“The students created a lot of great research and moved quickly; they came up with interesting and great solutions,” said Fitzsimmons, who earned a master’s degree in electrical and computer engineering in 2021 from IUPUI. “A lot has changed in the AI field in three years, and they documented these changes. Ultimately, we’ll be able to provide system users with actionable intelligence.”
“As AI/ML technology continues to be developed and deployed, we wanted to create some sort of visualization technique so that people at every level of [expertise] can find useful information and interact with AI systems,” he added. “If an adversary is trying to manipulate our systems, we wanted any user to be able to see that something was happening and know they can respond. A key output from IU built an open source graphical user interfaces for data visualizations so that users could better understand data distributions and potential anomalies.”
The partnership was also designed to introduce IU students to the wide variety of technical career paths available in the Navy.
Fiona Ryan, who graduated from IU Bloomington in 2020 with bachelor’s degrees in computer science and clarinet performance, was part of the NEEC project. Ryan, who is currently pursuing a Ph.D. at Georgia Tech and recently was awarded the National Science Foundation Graduate Research Fellowship, said her NEEC experience motivated her to continue to expand her knowledge of computer vision, a field of AI that trains computers to capture and interpret information from digital images, videos and other visual inputs.
“Having the opportunity to do undergraduate research on computer vision helped me learn how to build deep learning models and explore what state of the art models can and cannot do on difficult tasks,” Ryan said. “My work at Indiana University inspired me to continue pursuing computer vision research, now as a Ph.D. student at Georgia Tech. I am grateful to have had the opportunity to learn important research skills and formulate my interests in computer vision as an undergraduate student.”
Fitzsimmons said some of the IU students interned at NSWC Crane after participating in the program.
“Throughout the program, IU was engaged on the research side, students got involved, and there were several publications on the topic—which can then be applied to different Navy efforts for image-based ML analysis,” Fitzsimmons said. “A matrix of numbers isn’t helpful to users—this visualization will show users if the system is operationally appropriately or if it’s being attacked so they know what actions they can take. AI and ML is a technology area Crane is highly interested in, and this project ties in nicely with other ongoing efforts such as the Trusted AI research and workforce development initiative.”
Alicia Scott, a former chief engineer at NSWC Crane, said the research impacts other AI initiatives at NSWC Crane, including the recently launched Trusted AI project. Through the initiative, NSWC Crane has teamed up with IU, the University of Notre Dame and Purdue University to develop trusted AI research and workforce development that meeds the needs of the Department of Defense and the defense industrial base.
“NSWC Crane’s continued partnership with local universities in the realm of AI will help us not only enhance our impact for our own projects, but also the community as we continue to develop new methodologies and cutting-edge advancements in the big problems of trust in AI,” Scott said.
More information on this effort is available on the NSWC Crane website.
About NSWC Crane:
NSWC Crane is a naval laboratory and a field activity of Naval Sea Systems Command (NAVSEA) with mission areas in expeditionary warfare, strategic missions and electronic warfare. The warfare center is responsible for multi-domain, multi- spectral, full life cycle support of technologies and systems enhancing capability to today’s warfighter.