Hacker News new | ask | show | jobs
by direwolf20 133 days ago
Neural nets are structured as matrix multiplication, yet, they are universal approximators.
1 comments

You're missing the non-linear activations.