Skip to content ↓
Menu
What are you looking for?
An image of Carbin and Frankle at MIT
Michael Carbin (left) and Jonathan Frankle plan to explore why certain subnetworks are particularly adept at learning, and how to efficiently find these subnetworks.

Spotlight: May 7, 2019

To learn well, neural networks usually have to be large, with massive datasets. But what if they don’t actually have to be that big? CSAIL researchers have shown that far smaller subnetworks can be trained to make equally accurate predictions. Full story

May 7, 2019