Three University of Waterloo students have been awarded $17,000 after winning ClimateHack.AI, an international machine learning competition focused on climate solutions.
ClimateHack.AI is a student-focused AI competition that aims to reduce carbon emissions by developing more accurate techniques for forecasting site-level solar power production. Teams from 18 universities across the US, UK, and Canada were invited to compete and build machine learning models using data such as satellite imagery, weather forecasts and aerosol data. The research contributions of the top teams have the potential to be adopted by Open Climate Fix, a United Kingdom based non-profit, to produce more accurate solar power forecasts for the UK’s National Grid Electricity System Operator.
The Waterloo team featured students from the Faculties of Engineering and Mathematics. Carter Demars (BASc candidate, mechatronics engineering) Areel Khan (BMath candidate, mathematics) and Trevor Yu (MASc candidate, systems design) travelled to Harvard University this April to present their solution to machine learning researchers and industry experts. Team Waterloo impressed the judges with the most accurate model on the global leaderboard, despite being one of the smallest deep learning models submitted to the competition. “Our decision to ensemble small but powerful models allowed us to really outperform competitors, who often used models with significantly more parameters”, said Yu. Ultimately, the team took home the first-place prize of £10,000 (approximately $17,000 CAD).
The competition was also an opportunity to expand an important network of AI communities from institutions across the world. The team were able to explore Harvard campus, walk Boston’s Freedom Trail, and attend social events with the other competitors and organizers.
Jeremy Lo Ying Ping (Lead Organiser of ClimateHack.AI and Co-founder of DOXA AI), said: “We launched ClimateHack.AI as an initiative to provide talented students from around the world with the opportunity to meet and connect with one another – especially at our in-person competition finals – and to make a positive, real-world impact as an international community of machine learning enthusiasts by contributing to cutting-edge research aimed at tackling climate change.
“I am incredibly proud of what the team at the University of Waterloo have achieved together after many months of dedication and hard work towards solving this globally pressing challenge, and I look forward to seeing how their contributions go on to support the solar power nowcasting work of Open Climate Fix in the coming months.”
Demars, Khan and Yu will continue their work alongside Open Climate Fix (OCF), supporting the non-profit’s ambitious goal to reduce carbon emission by 100 kilo-tonnes per year through solar forecasting. The group is also actively involved with WAT.ai, a student-run Artificial Intelligence (AI) Organization at the University of Waterloo and a member of the Sedra Student Design Centre.
Trevor Yu, Areel Khan, Carter Demars
The Waterloo team present their solution.