What is mode collapse in GANs and how do you mitigate it?

hard

Answer

$Generator maps most latents to a few output modes because those fool the discriminator well — trained distribution becomes much less diverse than the data.$
$Symptoms: repeated outputs, poor coverage of the true distribution.$
$Fixes: minibatch discrimination (D sees a batch and can penalize identical outputs), unrolled GAN (train D a few extra steps), WGAN-GP loss for smoother gradients, spectral norm on D, feature matching, or diffusion models (which don't suffer from mode collapse).$

Check yourself — multiple choice

Discriminator failure only
Generator produces few outputs to fool D → poor diversity; fixes: minibatch discrimination, WGAN-GP, spectral norm, feature matching
Same as overfitting
Solved by more epochs

Mode collapse: G outputs limited modes; fixes involve richer signals or better losses (WGAN-GP).

#gan#generative

214 interview questions in this topic.