A Radical New Neural Network Design Could Overcome Big Challenges in AI

Researchers borrowed equations from calculus to redesign the core machinery of deep learning so it can model continuous processes like changes in health

An ordinary differential equation. Image: David Duvenaud

By Karen Hao

This story is for Medium members.

Medium curates expert stories from leading publishers exclusively for members (with no ads!). Register for a free account to begin your member preview.

Already have an account? Sign in

Join Medium to read this story.

Sign up with a free account to read this ad-free selection curated from MIT Technology Review.
Already have an account? Sign in