Crops News

Soy yield prediction model wins Syngenta AI Challenge


Syngenta and the AI for Good Foundation have named Tzvi Aviv, Ph.D., of Aviv Innovation and his partner, Vanessa Lundsgaard-Nielsen, Ph.D., of the University of Toronto, as the winners of the Syngenta AI Challenge. The contest recognizes and rewards innovative thinkers who use artificial intelligence (AI) tools to improve plant breeding by challenging them to create a model that predicts seed variety performance.

The team earned a $5,000 prize for its entry, Ensemble of Cubist Models for Soy Yield Prediction using Soil Features and Remote Sensing Variables, following finalists’ presentations at the SIGKDD Conference on Knowledge Discovery and Data Mining in Halifax, Nova Scotia. The winning submission set forth a novel mathematical model that can be used to help plant breeders determine the performance and viability of seed varieties—and inform commercial variety advancement decisions.

“The winning team’s entry demonstrated how effectively data helps solve the complex challenge of breeding high-yielding, geographically adapted seeds,” said Dan Dyer, Head of Seeds Development at Syngenta and lead for the AI Challenge committee. “All finalists displayed creative and analytical approaches that have the potential to benefit the agriculture industry.”

AI Challenge participants were provided access to a real-world data set and tasked to design a model to help scientists analyze extensive soybean seed data more efficiently and effectively, improving the plant breeding process and ultimately, the world’s ability to grow more crops with fewer resources. Due to the complexity inherent in agriculture, data driven models are increasingly being deployed in making seed breeding decisions.

“I am very interested in plant science and genetics, but I am also interested in doing something that is meaningful for me as a person,” said Aviv. “The contest was a fantastic opportunity for me to hone my skills around data analysis by working with the data, then working with models to identify which models work and which do not.”

The runners-up submissions were authored by teams representing Harvard University, Washington University in St. Louis, Carnegie Mellon University, and IN2 Group, a software company based in Croatia. Output was evaluated based on the submission’s scientific rigor, effectiveness in identifying the best varieties, and the finalists’ ability to clearly articulate their solution and its methodology.

Syngenta, an award-winning company for its innovation in plant breeding analytics, supported this competition, which was sponsored by the AI for Good Foundation and in partnership with IdeaConnection.

“The future availability of food is a global concern, and one that must be dealt with swiftly and sustainably,” said James Hodson, co-founder and CEO of the AI for Good Foundation. “The rapid adoption of artificial intelligence techniques across industries and growing skillsets in this field mean that the time to apply advanced analysis is now. We hope that competitions like this one will inspire emerging data scientists and researchers to help solve the world’s most pressing problems.”

Sponsored Content on AGDaily
Any views or opinions expressed in this article are those of the author and do not reflect those of AGDAILY. Comments on this article reflect the sole opinions of their writers.