AlphaGo is still based around layers of 3×3 local convolutions. That represents ... | Hacker News

Hacker Newsnew | past | comments | ask | show | jobs | submit

		waqf on Oct 18, 2017 \| parent \| context \| favorite \| on: AlphaGo Zero: Learning from scratch AlphaGo is still based around layers of 3×3 local convolutions. That represents a strong assumption about locality in the network design. I would expect AlphaGo to perform poorly on the game "Go with the vertices randomly permuted".

Palptine on Oct 19, 2017 [–]

Well they didn't use inception. If their inception units have, say, a 7x7 conv, then ladders will probably be found much earlier.

Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact