GAN: adversarial training, mode collapse and solutions
In 2014 Ian Goodfellow proposed a radically different idea: instead of defining a loss function by hand, train two neural networks to compete. One generates fake images; the other tries to catch fakes. The generator improves by fooling the discriminator; the discriminator improves by not being fooled. This adversarial game, when it works, produces sharper images than any prior method — but it is notoriously unstable.
Content is available with subscription.
Get full access to all courses on the platform for one year with a single payment.
▼
Unlike other platforms that charge per course, here you get everything for one price, and after one year of use there will be no automatic charge for the following year.