Allowing machine learning to ask questions can make it smarter

Biomedical engineers at Duke University have demonstrated a new method to significantly improve the effectiveness of machine learning models searching for new molecular therapeutics when using just a fraction of the available data. By working with an algorithm that actively identifies gaps in datasets, researchers can, in some cases, more than double their accuracy.

algorithm biomedical cases computer sciences data datasets duke engineers machine machine learning machine learning models questions researchers searching therapeutics university