Meet the DADAGAN
The DADA robot that draws
By Judy Mam
We are collaborating with artist and roboticist Alexander Reben to create a GAN (a Generative Adversarial Network) that learns how to draw from our data set of over 115,000 drawings.
We call it the DADAGAN. It sounds like a dragon out of a monster movie, but we like it.
This is a sample of the first batch of drawings generated by DADAGAN:
Funky robot art
DADAGAN at Tate Modern
We are thrilled to introduce DADAGAN at our live interactive performance at Tate Modern on November 29, thanks to our friends from MoCDA, the Museum of Contemporary Digital Art.
DADAGAN learns fast!
DADAGAN will generate the first drawing of Screens: An Exploration, the collaborative visual conversation we are creating for the event.
The world of Machine learning is both interesting, but also very scary. (For me personally, I'd say more scary.) I'm curious what type of parameters DADAGAN is set to so he doesn't go to far.
Either way, those are some interesting looking drawings. The robot is definitely learning how to draw fast.
@post-modern It is! What do you mean by too far? Alex Reben, the artist who is training it, is curating the drawings so obviously on this first batch, it learned faces pretty quickly. We want it to learn other shapes. But as we say in our blog post on the link, really it's the trainer who feeds the GAN and tweaks and curates. What is eerie is how it looks like it could be human made. Thanks for replying!
I was curious about what parameters the artist had so the robot doesn't disobey, nor do anything unwanted cause of an oversight in the programming. And those types of concerns didn't seem addressed in the article. I know next to nothing about machine learning, so I don't want to assume people like Alex Reben are careless right away, but I wanted to ask out of curiosity and personal concerns. One of my main thoughts is related to Isaac Asimov's rules for robots, particularly where they are supposed to obey their master.
I will have to check out the blog post linked later. It might go over some of my concerns that were not addressed reading this steemit post earlier this morning.
I will be gone for a while, so if you respond back, and don't see another response for a while, that's why. Thanks for your reply back, and have a great day!
@post-modern we are planning an in-depth interview with Alex Reben and will be sure to communicate your fascinating questions, because we are not roboticists ourselves. I'm not sure that the GAN can disobey, given that it's a system that just discriminates from a training dataset it is fed. It doesn't have volition. It may make mistakes of logic.
For more info follow our blog
Oh cool!
Thanks for the response and verification.
I'll plan on following the blog. It sounds quite interesting. Thanks!
Congratulations @dadanyc! You have completed the following achievement on the Steem blockchain and have been rewarded with new badge(s) :
You can view your badges on your Steem Board and compare to others on the Steem Ranking
If you no longer want to receive notifications, reply to this comment with the word
STOPTo support your work, I also upvoted your post!
Vote for @Steemitboard as a witness to get one more award and increased upvotes!