An idea that combined solutions from multiple AI models to predict wildfire growth took the top prize at the first hackathon at the University of Waterloo intended to identify cutting-edge solutions for wildfire management.
The winner was one of 27 teams of Waterloo students who presented their ideas to a panel of industry leaders, faculty and students. The Waterloo Data and Artificial Intelligence Institute, WAT.ai, a student AI organization, and the UWaterloo Data Science Club hosted the event on May 25.
Canada’s 2023 wildfire season was the most destructive and costliest on record with more than 6,500 fires burning across the country last year.
“Wildfires have a direct and measurable impact on our quality of life and on the state of our natural environment,” said Carter Demars, an undergraduate student in the Department of Mechanical and Mechatronics Engineering and member of the winning team. “It’s incredibly exciting to have our work recognized by wildfire experts for its potential in wildfire forecasting and intervention.” The other students on the first-place team—called We Didn’t Start the Fire—include Trevor Yu, from systems design engineering, Areel Khan, from mathematics, Ken Wu, from computer science and business administration, and Kiernan McGuigan, from systems design engineering.
More than 130 students from all six Faculties at Waterloo participated in this event.
“This hackathon is more about collaboration than just competition,” said Dr. Stephen Smith, a professor in the Department of Electrical and Computer Engineering and co-director of the Waterloo Data and Artificial Intelligence Institute. “It’s about working together to address a complex and urgent issue. It’s about pushing the boundaries and making a real impact on wildfire management strategies.”
First prize was $1,000 and merchandise. The second-place team share $500 and there was a two-way tie for third and $250.
“Beyond just technical solutions, the top teams were able to demonstrate the importance of interdisciplinary thinking by applying domain knowledge on wildfires about weather, topography and forestry to further improve technically sound solutions,” said Bing Hu, co-founder of WAT.AI and a PhD student in the Cheriton School of Computer Science. “Combining smaller models in a powerful ensemble approach is one of the best ways to creating highly scalable solutions, which is a critical element to real-world applications of wildfire AI.”
MAG Aerospace, Rogers Communications and Natural Resources Canada sponsored the event.