New research challenges a common idea of how ML algorithms "think" about certain tasks

There’s a long-standing idea that as raw inputs get transformed to these intermediate representations, the system is trading prediction for compression, and building higher-level concepts through this information bottleneck.

Read more at https://www.technoscious.com/home/new-research-challenges-a-common-idea-of-how-ml-algorithms-think-about-certain-tasks