Paul Vicol, NSERC USRA Scholarship Winner
Paul Vicol is living out his childhood fantasy. He received two NSERC Undergraduate Student Research Awards (USRA), each valued at $4,500, for two semesters of research studying belief change, a subfield of artificial intelligence (AI), under the supervision of SFU professor James Delgrande.
He explains his research, passionately scribbling examples on a whiteboard. “We have a group of agents, each with some initial beliefs about the world,” he says. “The agents communicate with each other: they share their beliefs, and learn from other agents. Our goal is to find out what each agent believes after learning as much as possible from other agents.”
In the original problem, which he worked on in 2013, agents do not give up their original beliefs, because those beliefs represent first-hand experience, and are considered more reliable than beliefs from other agents. So, if one agent believes that the grass is blue, and another believes that the grass is green, neither will change its beliefs. Their own experience trumps what they hear from others.
“This summer I am looking at ways to extend the framework by adding the notion of expertise,” he explains. Agents may have different levels of expertise in certain beliefs, representing how "sure" they are about those beliefs. With this addition, an expert can convince an agent to give up its original beliefs. “So now, if one agent believes that the grass is blue, but another agent with greater expertise believes that the grass is green, then the first agent will give up its original belief, and incorporate the belief that the grass is green,” he says.
He is also creating a user interface for the system to allow people to experiment and see how beliefs change in different scenarios.
Vicol is one of the first people to analyze belief change operations and their importance in real-life applications. This fascinating field of research hints at the possibilities of a world where robots can make their own decisions and autonomously navigate public spaces.
“Imagine a group of robots where each has a sensor and each gets different information about the environment,” says Vicol. “This information is limited, but by belief merging and sharing, each robot can communicate with the others and form a more accurate picture of the environment.”
Vicol’s research could be used to enable robots to assess dangerous situations. For example, firefighters could dispatch a team of robots into a burning building to map the layout of the room and plan the safest approach.
The technology could also be applied to weather sensing systems to obtain data for locations in-between weather stations.
Vicol enjoys the self-directed research offered by the USRA and the opportunity to explore ways of approaching open problems that have never been tackled before. The third-year student, who hopes to pursue graduate studies in artificial intelligence or networking, meets with Delgrande every week to solidify his objectives and goals.
“The NSERC USRA is a really unique opportunity for undergraduate students to get research experience,” he says. “You get the freedom to explore different topics and prove you have research potential before you even go to grad school. You get to work on one topic in-depth over a long period of time and it’s great preparation for future research projects.”
Update: Paul Vicol was awarded the 2015 Governor General’s Silver Medal for academic excellence and is now master’s student at SFU, continuing his research into belief change with professor Delgrande. He has established the Paul Vicol Award for Computing Science Research ($1000), to be awarded for the first time in Spring 2016.