Generative Adversarial Nets (GAN) are tools for generative modeling using deep learning techniques, such as in deep convolutional neural networks. GAN is a generative model that is trained using two neural networks. One model is called a “generator” or “generative network” model, which is trained to generate new probable patterns. Another model is called a “discriminator” or “discriminator network” and learns to differentiate generated examples from real examples. They study at the same time. A generative network generates a fake, and a discriminative neural network determines how realistic the generated sample is by comparing it with the real sample. Generative and discriminative neural networks behave like opponents in some game, hence the name adversarial neural networks. GANs are capable of realistically generating new human faces, as well as altering images by coloring photos, aging faces using high resolution.
You can also create new animated characters or a new logo. Modeling new clothes or samples of new furniture can also be generated using GANs. Generative adversarial networks are used to create object samples for scenes in computer games. These networks are known to be used by Facebook. Generative adversarial networks are a very young field, the first article about such networks appeared in 2014. Using generative adversarial networks, we can learn to create realistic fake versions of almost anything. The article includes a collection of sites that have appeared over the past month.
The site from which it all began, the name of which speaks for itself. Built using a generative adversarial network (StyleGAN), this website created excitement and intrigue in the tech community and inspired many other sites.
By drawing a beautiful parallel between the generators and discriminators in GANs and the apprentices and potters, this site demonstrates the remarkable ability of neural networks to exquisitely replicate the mastery of professionals in yet another craft.
Now we can generate lyrics for a song on a specific topic. If only we could combine this with a text-to-speech and melody generation model to create completely original songs. Then the music industry would take GAN seriously.
By creating realistic Stack Overflow questions, you can keep them open on your computer so it feels like you’re actually doing the work. There is also a “Other random questions” section on the right side of the page that shows other equally synthetic questions.
To showcase its resume template builder, TextgenRNN, a recurrent neural network (RNN) specialized for text generation, is included. Bonus points – the headshots used in each resume are from thispersondoesnotexist.com.
Honestly, if I had to quantify the site with the greatest added value to humanity, it would be in the top three. Loads of memes and more on every page load. You can hang for more than one hour, we guarantee. Very interesting service.
If your IP address is in San Francisco, this site is actually just redirecting to a random portfolio company from an AI, blockchain, or crypto-focused venture capital firm.
Why try to find the perfect home when you can create one instead? Just find a place you like or have a good climate, buy land, build a house and relax and enjoy the rest of your life.
By using GPT-2, this website manages to generate words that sound like they should exist, but don’t. Great for startup names or baby names (now that the coveted X Æ A-12 is taken).
Before I explain what this site is about, you can sit down, preferably in one of the chairs created by this GAN. What’s more, all the chairs are 3D and you can fine-tune the “weirdness”.
Note that this is an SMS chatbot. You can send a message to get photos of the feet. Pictures are animated. There are no legs. Why do you want to do this? Who knows.
This site demonstrates the power of a relatively simple model capable of interpreting emotions from facial expressions. The results may shock you. Or please you. Or make you angry.
This site is a little weirder than the others, but you’ll read more of the story than you expect. It’s very exciting.
With a website, you can get a unique new look at the top every time you refresh the page. Very useful and convenient.
Elon is said to have used this website to develop the Cybertruck. Thanks to the site, he made some small changes to the curvature of the left windows.
With the state of politics in the world, who wouldn’t want a new face at the helm? The author of this site created a GAN on real faces of the UK Parliament. But the faces created by this model don’t really exist, but they still get my vote.
It’s not hard to see why this site is influential. In fact, while you are looking at this site, you can see that the site is looking at you. A useful feature would be Sephora links to the makeup featured, but I wouldn’t expect this feature to appear anytime soon.
Do you need a tropical vacation? We have the perfect offer for you. Exclusively through its website, Vojtěch can sell you a rental property on the beach without tourists! Of course, the property only exists inside the computer, but you’re paying for the experience.
I have seen the dark side of the interwebs and it is 100% darkness. Visit this website at your own risk and you will never watch The Simpsons again.
On this site, you can pan the generated ponies, zoom in, and even auto-expand on hover. You have no idea how convenient it is.
This site presents a grid of AI-generated fur portraits trained by arfa using the StyleGAN2 architecture. Very useful and interesting information.
Combining OpenAI’s Jukebox and StyleGAN for visuals, this website brings you a never-before-seen VMA-worthy music video. Watch out MTV – you’re going to have a lot more nominations this year.
Want to surprise someone special with a new recipe? Yes, this is not the site for you. But if you want to share a new AI-powered site that boasts GAN excellence, this is the site for you!
It’s not often that a country wants a new flag, but if they do, you’ll find them on this site. The site will suggest a potential new name for the flag. Just in case they want to change the name too.
Looking for an interesting new side project? Look no further, this is a new site that recommends innovative new ideas for open source companies or projects.
Watch out for your sneakers, this site will blow your mind. Feature Wishes: A button on each sneaker that would instantly send a photo of the sneaker to Nike.
Fortunately for MLB, we’ve developed a fantastic system for recruiting new baseball talent – create them with the help of AI and you’ll have your own talent.
As Apple Maps and Google Maps compete to improve the quality and detail of their maps using satellite imagery and artificial intelligence, this site takes a very different approach – simply create your own maps with as much detail as possible.
More than 55% of the world’s population lives in cities with a huge amount of lighting, not having the opportunity to see the beauty of the stars and constellations. Without the opportunity to see the wonders of space.
On this site, the actual voices sound completely realistic, but the generated sentences are absurd. Voices may not pass the Turing Test, but they’ll make you wonder who’s really on the other end of the phone the next time you call support.