Generative Adversarial Networks (GANs): A Simple Explanation

Adithya Thatipalli
4 min readJul 11, 2023

Have you ever played a game of “two truths and a lie,” where your friend tells you two true things and one false thing, and you have to figure out which one is the lie? Well, there’s a computer program that plays a similar game, but instead of telling lies and truths, it creates and detects real and fake data. This program is called a Generative Adversarial Network, or GAN for short.

What is a GAN?

A GAN is a pair of computer programs that work together, just like a team. But there’s a twist: they’re also kind of working against each other, like two players in a game. This team has two members: the “generator” and the “discriminator.”

The Artistic Generator

Let’s imagine the generator as an aspiring artist. This artist wants to create beautiful paintings that look like they were painted by famous artists. The generator’s job is to create (or generate) new data that looks exactly like the real thing it has seen before.

In the world of GANs, this could be anything: a picture of a cat, a piece of music, even a video. The generator uses all its knowledge and skills (which are actually complicated math and algorithms!) to make these creations as believable as possible.

The Detective Discriminator

Now, let’s imagine the discriminator as a sharp-eyed detective. This detective is an expert in art and can usually tell if a painting is a genuine masterpiece or a copy made by our generator artist.

The discriminator’s job is to look at both the real data (the original masterpiece) and the fake data (the painting our generator made) and figure out which is real, and which is fake.

The Exciting Game Between Generator and Discriminator

Every time the generator creates a new painting, it shows it to the discriminator. If the discriminator can spot it’s a fake, it gives feedback to the generator, like a friend telling you why your lie in “two truths and a lie” wasn’t believable. The generator then uses this feedback to improve its next creation.

On the other hand, if the discriminator can’t tell the difference between the fake and the real one, that means the generator has done a great job! The generator has “fooled” the discriminator successfully.

This back-and-forth game continues, with the generator always trying to create better paintings and the discriminator always trying to get better at telling them apart. They are “adversarial,” which means they’re sort of like opponents in a game, but they’re also helping each other improve.

Why are GANs Important?

GANs are magical because they can create things that didn’t exist before, like a new image of a cat that doesn’t look like any real cat but still looks like a cat. They are used in many areas, from creating realistic video game graphics to designing new types of clothing in fashion.

However, just like any powerful tool, GANs can also be used in wrong ways, such as creating fake videos of real people saying things they never said. That’s why it’s important to keep learning about GANs and understand how they work, so we can use them for good and also be aware when they might be used in not-so-good ways.

In the end, Generative Adversarial Networks (GANs) are a fascinating technology that combines creativity and detective work, making computer programs that can create and discern between real and fake data. And even though it’s a complicated concept, kids like you can understand it, too! So, keep asking questions and keep learning.

Generative Adversarial Networks (GANs) have been used in various ways in different products and services. Here are a few examples:

1. DeepArt and DeepDream: These are tools that use GANs to transform your photos into artwork, in the style of famous painters, or to generate unusual and imaginative images from your photos.

2. Deepfake Technology: This is a controversial use of GANs, but it’s one of the most known. Deepfake technology can generate realistic fake videos and images of people, often used in movies or TV shows to create special effects. But it can also be misused to create false content, which can be misleading or harmful.

3. This Person Does Not Exist: This is a website that uses GANs to generate images of faces of people who do not exist. Each time you refresh the website, a new face is generated. It’s a fascinating demonstration of how powerful GANs can be.

4. AI Painting: There are companies, like Promethean AI, that use GANs to assist artists in creating artwork, particularly in the video game and movie industry. The AI can generate images such as landscapes, characters, and other objects, based on a simple description.

5. ChatGPT: This is a model developed by OpenAI that generates human-like text based on the input it receives. While not strictly a GAN, it’s a related concept and shows how these generative models can create very realistic output.

6. Fashion Design: Some companies are using GANs to create new fashion designs. The AI can generate images of new types of clothing, allowing designers to gain inspiration from these AI-created designs.

Remember, as exciting as this technology is, it’s also important to consider ethical considerations because GANs can be used to create misleading or false information. It’s crucial to use these technologies responsibly.

More content at PlainEnglish.io.

Sign up for our free weekly newsletter. Follow us on Twitter, LinkedIn, YouTube, and Discord.

--

--

Adithya Thatipalli

Security Engineer by Day, Cloud and Blockchain Learner during Night