Stevens Gets Real About Artificial Intelligence
The university leverages AI to make our lives healthier, safer and more secure in a variety of ways.
Artificial intelligence, or AI, is already shaping many aspects of modern life. Siri and Echo use AI-type processes; so do Google Maps, Google Translate and Amazon’s recommendation engine. IBM uses it to predict the weather. It helps your phone recognize your face.
The next waves of cars, hospitals and manufacturing plants will almost certainly incorporate these technologies.
Now Stevens is a growing player in the new move to harness AI for societal benefit. The Stevens Institute of Artificial Intelligence (SIAI) launched in 2018, even as researchers across campus deploy AI and learning technologies in disciplines ranging from medicine and security to emergency planning and sports science.
“Our vision is to drive AI research and application that solve some of those tough ‘big’ problems that have so far resisted solving,” explains K.P. “Suba” Subbalakshmi, SIAI founding director.
Subbalakshmi and colleague “Mouli” Chandramouli have leveraged AI. to produce a series of remarkable innovations tackling everything from elder care to banking fraud. Their work began with a set of algorithms that accurately detect lies in writing or speech. Next the duo trained software to scan texts and phone calls for early-warning symptoms of Alzheimer’s disease, dementia and aphasia. (A doctoral student later created an application based on the tech that tests cognitive abilities using a smartphone.) Still another tool developed by the pair uses AI. to spot signs of insider trading or other fraud by picking out unusual patterns in speech, the use of code words and other signals.
A group headed by computer science chair Giuseppe Ateniese designed an AI network that quickly became skilled at guessing private pictures from a locked phone without even touching it. (He did it to create awareness about the security and privacy risks of AI)
Researcher Samantha Kleinberg develops models and methods aimed at improving health, including tools that model uncertainty, factor in missing data and enhance the analysis of medical data. The research is applicable in treating stroke and diabetes patients, among others.
Two business students shared top prize in a UBS-sponsored competition requiring participants to develop and present AI-driven technologies that optimize the selection of wealth management branch offices. Dean Kelland Thomas harnessed AI to teach computers the tendencies of jazz masters and improvise in real time with live players; the work may have applications for defense, computing and virtual assistant design.
Stevens researchers have also used machine learning to guide robots in tasks such as underwater navigation (led by Brendan Englot) and the evacuation of public spaces during emergencies (led by Yi Guo).
And there’s a growing effort in the exploding field of computer vision. Projects include work by Xinchao Wang to analyze transit, medical centers and sports videos and professor Philippos Mordohai’s work to improve motorized wheelchairs and augmented reality tech.
“We have such breadth,” sums up Subbalakshmi. “It’s really our strength.”