College of Engineering News Room
The AI+X Institute: a university-wide asset for USF researchers working with artificial intelligence
by Russell Nay
During its first official event with external participation in February 2020, (AI+X) welcomed Amazon vice president/distinguished scientist and University of Southern California professor GĂ©rard Medioni to talk about his 40-year career in the field of computer vision research.
After , AI+X institute affiliates had smaller group and one-on-one discussions with Medioni, who joined Amazon in 2014 to lead the development of technology used in A and later . Since then, the institute has held five major events featuring notable industry researchers in the field of artificial intelligence, a bootcamp on deep learning methods, and weekly seminars in the fall and spring semesters featuring the research of faculty members from colleges across USF who work in artificial intelligence.
Co-directed by Department of Computer Science and Engineering professors Larry Hall and Sudeep Sarkar, the AI+X institute is a university-wide research and education center that aims to establish a world-class academic research and development center at USF. Its members will conduct externally-funded research in AI and work with industry partners in healthcare, finance, manufacturing, cybersecurity, biology, and transportation to ethically and responsibly transition advances into beneficial products.
“Companies are reaching out to us to help them out with problems that they’re facing,” Sarkar said.
The institute is conducting research with a number of AI corporations, including TheIncLab, an AI and data visualization company that recently relocated to Ybor City from Washington D.C. , Tampa Mayor Jane Castor said that USF plays a critical role in Tampa’s emergence as a magnet for entrepreneurs and innovative business startups and expansions.
“The USF talent pipeline and cutting-edge expertise in bio/life sciences, technology, cybersecurity and so much more are central parts of Tampa Bay’s appeal to business leaders across the globe,” Castor said.
The AI+X institute was founded in July 2019 following a $120,000 ßŮßÇÂţ» Strategic Investment Pool funding selection by university leadership and the Florida Board of Governors. The institute began with more than 20 affiliated USF faculty members from the colleges of Engineering, Arts and Sciences, Medicine, and Business. By 2020, it grew to involve over 110 affiliated faculty members from a variety of colleges at USF and received more than $1.7 million from research proposals with organizations including the National Science Foundation, the U.S. Department of Defense, and the National Institutes of Health.
Sarkar said that the concept of the institute was received well by the university research and innovation review committee involved in the funding process.
“In the first year itself, the university’s return on investment was more than 10 times,” he said. “We’re looking forward to putting more research proposals together in the future.”
The institute will also offer graduate students a certificate in AI as well as classes for undergraduate and graduate students offered through the Department of Computer Science and Engineering. The work of many institute collaborators was showcased in the February 2020 , including those in USF Health as well as in the colleges of Engineering, Medicine, and Business.
Department of Industrial and Management Systems Engineering associate professor Susana Lai-Yuen said that her involvement with AI+X events and co-collaborators led to establishing connections that funded her research and later helped expand it.
“The institute has definitely helped me in establishing connections with other researchers, universities and funding agencies,” Lai-Yuen said. “(My research partners and myself) were recently awarded a research grant from the U.S. Department of Defense. The hope is that these connections will lead to other projects and external funds.”
Lai-Yuen develops optimization methods for . This work can be used to automate the identification and extraction of data from different types of images.
She and several other AI+X affiliated faculty members said they’d like to see an expansion of the size and strength of the institute’s GPU cluster and other computing resources provided through the Department of Computer Science and Engineering.
Department assistant professor Shaun Canavan said that he and his research partners had their research funded through AI+X institute connections too. They use AI to , which has applications in a variety of industries.
“The institute’s been a huge boost to my career,” he said. “Some of my largest grants came from working with (institute-affiliated) faculty, and I use the computer cluster every day. Without that cluster, we couldn’t do our work.”
Civil and environmental engineering associate professor and Susan A. Bracken Faculty Fellow Xiaopeng Li also makes frequent use of this cluster. While researching how can improve the mobility, sustainability and safety of transportation systems through smart vehicles connected to each other and roadways, Li uses data models and processes vehicle sensing and image data.
He said that connecting vehicles with a more powerful type of cluster could allow him and his team to one day turn the USF Tampa campus into a living lab for their research that connects information on pedestrians, cars, buses, traffic flows and potential road hazards in real time.
“A very nice thing about the AI+X institute is that it’s a platform for exchanges, communications, and discussions of possible collaboration across multidisciplinary faculty,” he said. “Past interactions led to eventually getting a multi million-dollar grant from the Department of Energy.”
Department of Computer Science and Engineering assistant professor John Licato researches the in his lab and said that the institute’s aligning of researchers with a focus on AI presents a valuable opportunity for USF faculty members to potentially receive more funding from industry and government sources.
“There’s often a really difficult alignment problem where you have all these funding opportunities, especially in AI, which often have very specific goals that make it hard to figure out which ones will meet the interests of both the researcher and the opportunity,” Licato said. “If you have this central institute that serves as conduit and speaks to both sides, then that alignment problem gets better. It’s something I’m already seeing the start of.”
He added that this alignment of researchers with a common interest in AI also allows for a more efficient use of resources.
“In one of my sub fields, natural language processing, computing is really expensive,” he said. “You have these million-dollar clusters that still aren’t powerful enough to do all the things researchers want to do. Individual professors could never get the funding by themselves to carry this out practically. But if there were a program in the institute where they were able to pool resources together for professors with common goals, that’s a huge potential benefit.”
Department of Industrial and Management Systems Engineering assistant professor Ankit Shah’s encompass the detection and mitigation of a wide range of physical and digital threats to defense systems. Shah develops AI-enabled decision support frameworks to enable intrusion detection systems with human operators in the loop in making better decisions under uncertainty.
Shah said that past AI+X seminars on deep reinforcement learning and computer vision were very helpful for his area of research, and he’d like to see an emphasis on industry connections in future institute events.
“It’d be very useful to have a program that matches researchers in the institute with the appropriate industry connections based on their disciplines and domains,” he said. “It would be advantageous to understand and cater to the various problems faced by industry partners.”
Hall said that goals set for the AI+X institute’s future currently include securing long-term, external funding, refreshing the department’s GPU cluster and supporting equipment, and deciding on the metrics needed to continue building a world-class research reputation for the institute.
“Ideally, we would have perhaps one of the NSF research institutes that we’ve looked at, on board,” he said. “In general, a pipeline to success may look like a number of successful faculty affiliated with the institute securing X amount of dollars in funding and publishing Y number of high quality papers each year.”