Sign In

Communications of the ACM

ACM TechNews

Incorporating Human Error into Machine Learning

View as: Print Mobile App Share:
A doctor reviews brain scan images.

The researchers found that training with uncertain labels can improve these systems’ performance in handling uncertain feedback.

Credit: PeopleImages/Getty Images

Scientists at the U.K.'s University of Cambridge, The Alan Turing Institute, Princeton University, and Google DeepMind are incorporating uncertainty into machine learning (ML) systems

The researchers utilized established image classification datasets so humans could supply feedback and rate their uncertainty level when annotating specific images.

They learned the systems can handle uncertain feedback better when training with uncertain labels, although their overall performance degrades rapidly with human feedback.

Cambridge's Matthew Barker said, "We're trying to bridge [behavioral research and ML] so that machine learning can start to deal with human uncertainty where humans are part of the system."

From University of Cambridge (U.K.)
View Full Article


Abstracts Copyright © 2023 SmithBucklin, Washington, D.C., USA


No entries found