HN
Hacker News
Top
New
Best
Ask
Show
Jobs
Comment by bilsbie | Hacker News Reader
Full thread
bilsbie
·
Do they use gradient descent? Otherwise how do they train?
View on HN
bhaney
·
From the paper, it looks like they apply a long random stimulus when it misses the ball, and a short predictable stimulus when it doesn't. The theory is that the neurons will try to organize in a way that avoids the random stimulus.
Lapsa
·
cool. tldr appreciated
Reply on news.ycombinator.com