GAN
A GAN (Generative Adversarial Network) is a pair of neural networks trained against each other: a generator creates synthetic samples while a discriminator judges them against real data, and each improves by trying to beat the other.
No jargon, no math
In plain words, what is it?
Two players, one table. The generator takes random noise and turns it into a fake sample; the discriminator sees real and fake samples mixed together and must say which is which. Every verdict becomes feedback: the discriminator learns to catch subtler fakes, the generator learns to produce more convincing ones. They level up together, like a forger and a detective training on each other. The equilibrium arrives when the fakes pass as real. The known weak points: if the generator finds one type of fake that always works, it stops varying (called mode collapse), and training can oscillate for a long time before settling, or never settle at all.
- 01
Sample. Random noise goes in as the seed for a new synthetic record.
- 02
Generate. The generator turns the noise into a fake.
- 03
Judge. The discriminator scores the fake against real data: real or fake.
- 04
Feedback. The generator adjusts to fool the judge better next round.
- 05
Repeat. Rounds continue until the fakes pass as real.
Terms that live nearby.
Generative AI
Generative AI is software that produces new content, text, images, audio, code, by sampling patterns learned from vast training data, rather than only classifying or retrieving what already exists.
ReadDiffusion Model
A diffusion model is a generative model trained to reverse a gradual noising process: it learns to remove noise step by step, so at generation time it can start from pure static and refine it into an image, sound or other output that matches your prompt.
ReadSynthetic Data
Synthetic data is artificial data engineered to reproduce the statistical patterns of real data: a model or a set of rules generates new records that behave like the originals without containing any real person's or customer's information.
Read§ 04
Want this running in your stack?
We design and ship AI systems with permissions, logs and human approval. It starts with a 30-minute scoping call.