Neural Slime Volleyball
Games
neural network game
SHOW HN2015
How HN took it
13 positive, 26 neutral or asking, 1 mixed, 6 critical, out of 46 comments about the project. Thread score +0.06.
Idea: 8 for, 2 againstDesign: 0 for, 3 againstDocs: 1 for, 0 againstPerformance: 0 for, 1 againstExecution: 1 for, 0 against
The thread, comment by comment
Each row is a top-level comment and its replies, in HN's order. Select a cell to open that comment on HN.
positive neutral or mixed critical about another commenter or something else the maker
What HN said
What people liked
Interesting neuroevolution approach
From 2 comments
[accompanying blog post](http://blog.otoro.net/2015/03/28/neural-slime-volleyball/) I too am perplexed as to what the neurons represent, the blog describes the RNN as 19 inputs…
deepnet on HNWhile Deep-Q-Learning is powerful, it uses backprop - your use of CNE genetic evolution rather than backprop may provide a more global search than Mnih's Deep-Q. The methods seem…
deepnet on HN
Fun old-school game
From 2 comments
Yeah, judging the angles and getting the ball to bounce in the right direction is really hard. But I'm gonna see if my kids can play this. I'm guessing they probably can. I like…
uuave on HN[accompanying blog post](http://blog.otoro.net/2015/03/28/neural-slime-volleyball/) I too am perplexed as to what the neurons represent, the blog describes the RNN as 19 inputs…
deepnet on HN
Objections
Game is very hard to play as a human
From 2 comments
This is cute but I think it's just a really hard game to play as a human. Your little dude can't move fast enough.
ebbv on HNThis is massively frustrating :) although I started to develop some strategies once I got used to the controls. I have a question though: In useful.js, there's a function like…
chillingeffect on HN
Arrow key bindings fail on non-QWERTY keyboards
From 1 comment
Please modify the bindings, the directional keys are international, without a qwerty keyboard I can't play - and no, I will not modify it systemwide for just a JS game. Nice one…
data_scientist on HN
Requests
Version that starts untrained so training can be watched
From 2 comments
Hope it didn't seem like I was trying to criticize. The app is really cool, and thanks for open-sourcing it. I'd be really interested in seeing the version that doesn't start…
squeaky-clean on HNAs requested, I put up a demo of neural slime volleyball without any pre-trained agents so you can see them improve gradually as they evolve:…
hardmaru on HN
Write-up explaining how it works
From 1 comment
Whoa.... any chance you can write up something on how this works? I've never seen a working implementation of a recurrent net before. I would be very grateful!! Either way, this…
cowpig on HN
Questions
How the network was trained
From 1 comment
The interesting thing, to me, is how was network the trained?
discardorama on HN
Whether it keeps learning from the human player
From 1 comment
Does it continue to learn when I take over control, eg: learn from me, or is the learning done at the time I can play?
donatj on HN
Alternatives named
Original Java Slime Volleyball game
From 1 comment
Very interesting-- slime volleyball was quite the thing back in my high school days. (See it here if you can get the Java applets to run in 2015 [1]) I was surprised to see a few…
bufordsharkley on HN
Schmidhuber's Evolino
From 1 comment
The method is not unique. See some of the work of Schmidhuber's group. They are doing a lot of reinforcement learning for recurrent nets (LSTMs) and also via evolutionary…
albertzeyer on HN