Skip to content

Comment on General Theory of Neural Networksparent

Comments

The topology needs to be information bearing. Weights of 0.0001 are likely spurious and if other weights are so relatively big they can effectively make the other fan in weights spurious as well.

AboutSource Built by g1lg1l

Hackerly is an independent reader for Hacker News, built on the public HN API. Not affiliated with Y Combinator.