Magic + AI = mandatory upvote. Nice one!
Generating Magic cards using deep, recursive neural networks
61–69 of 69 posts
Re: Generating Magic cards using deep, recursive neural networks
#62Magic + AI = mandatory upvote. Nice one!
I am enamoured that geeks can find amazement in computer generated content. I wonder if Shakespeare scholars felt the same with the recent neural net creating Shakespeare-like content. What is to come, when cgc starts to become even more amazing? A individualized, computer generated TV-shows that hit all your favorite subjects?
Re: Generating Magic cards using deep, recursive neural networks
#63Would be interesting to do this for other other games (that interest me more ;)).
Netrunner ICE generation would be interesting. I'd also like to run some algorithm over all Star Wars LCG cards and have it generate new arrangements of objective sets. [for those that don't know the game you don't build your deck by picking cards like in Magic but rather predefined bundles of 6 cards (objective sets)]
Re: Generating Magic cards using deep, recursive neural networks
#64Wonder what MaRo and the gang thinks about this. I am pretty surprised at the results so far from that post. They mostly don't make sense but I'm sure if you went through say 100 of them you're bound to find one or two legitimate ideas in there.
Unfortunately they might not be able to view this - as it might be viewed as "unsolicited design" - which they can't view due to company policy.
Stops people suing Wizards if they print a card identical or similar to a homebrew card.
Re: Generating Magic cards using deep, recursive neural networks
#65Earlier quoted context omitted.
I'm not so sure this card is on cost. My neural network is telling me that it's actually too good at 4 mana. Don't forget about potential combos like having a bunch of zero cost creatures in your graveyard + anything else that triggers when you play a spell or when a creature enters the battlefield. But even without that, this is a card advantage machine.
It's rather good at 4 mana, though that could be balanced in the metagame by making it rare or mythic. What zero-cost creatures did you have in mind? Those aren't incredibly common. And in any case, a combo that requires 3-4 specific cards is not overpowered; one that only requires 2 specific cards often is.
You cannot balance a card [in constructed] by making it rarer. You just make it so fewer people can run the deck based around it in a given tournament.
It wouldn't be particularly broken in draft anyway - which is the only meta where rarity does impact balance to some extent.
Re: Generating Magic cards using deep, recursive neural networks
#66How about IDE that suggest you next line of code? Or next variable, or method to call? Add static analysis and it can actually be fairly useful. E.g. val name = object.[maybe you wwant to use getFirstName]?
Believe me, it's been thought of (it's been in my head ever since my colleague Frank Seide introduced me to RNNs 4 years ago, I'm sure I wasn't the first!). It is one thing to generate code that looks like code, quite another to generate code that is useful. Character-by-character probably isn't good enough for this task. Anyways, I talk about how to use probability (but not compute it!) in http://research.microsoft.…
The old Jungloid technique for Java factory automation/type conversions (I have an A, I want a B) seems like a possible use for this kind of thing. http://dl.acm.org/citation.cfm?id=1065018
Re: Generating Magic cards using deep, recursive neural networks
#67Earlier quoted context omitted.
Believe me, it's been thought of (it's been in my head ever since my colleague Frank Seide introduced me to RNNs 4 years ago, I'm sure I wasn't the first!). It is one thing to generate code that looks like code, quite another to generate code that is useful. Character-by-character probably isn't good enough for this task. Anyways, I talk about how to use probability (but not compute it!) in http://research.microsoft.…
The old Jungloid technique for Java factory automation/type conversions (I have an A, I want a B) seems like a possible use for this kind of thing. http://dl.acm.org/citation.cfm?id=1065018
There has been a lot of work on code mining. It still isn't that useful, and no one has used RNNs (well, there is one paper I remember reading).
Re: Generating Magic cards using deep, recursive neural networks
#68Earlier quoted context omitted.
It's rather good at 4 mana, though that could be balanced in the metagame by making it rare or mythic. What zero-cost creatures did you have in mind? Those aren't incredibly common. And in any case, a combo that requires 3-4 specific cards is not overpowered; one that only requires 2 specific cards often is.
you need two zero cost creatures and a way to sacrifice them for benefits, then just trade off. I can think of infinite mana with four cards.
Yep, Ashnod's Altar, that card, and any 1-2 cost artifact creatures and you're set.
FWIW, it's about the same cost as Enduring Renewal.
Re: Generating Magic cards using deep, recursive neural networks
#69I was very excited to learn about recursive neural networks and how they are different from recurrent neural networks. I began imagining self-similar fractal topologies and automatic convolution layer creation etc... Disappointed to see it was just a typo :-)
There are actually recursive neural networks. https://en.wikipedia.org/wiki/Recursive_neural_network