U of T Engineering study finds bigger datasets might not always be better for AI models
New research on materials science datasets shows the amount of training data can be significantly reduced with minimal impact on the performance of the model