Loading your learn pack…
Read the summary, then start training!
GANs can be defined as a two-player zero-sum minimax game between a generator (minimizer) and a discriminator (maximizer).
In a GAN, the generator and discriminator are framed as two players in a zero-sum game: the generator tries to produce samples that look real, while the discriminator tries to distinguish real data from generated data. Formally, the original GAN defines a game between a generator strategy (a probability measure over the data space) and a discriminator strategy (a Markov kernel mapping inputs to probabilities in [0,1]). The objective function L(μG, μD) is constructed so that the generator minimizes L while the discriminator maximizes it, making one player’s gain the other’s loss. In practice, this game is implemented by alternating updates: the discriminator is trained on real samples and generator-produced samples to improve its ability to label inputs as real vs. fake, while the generator is trained based on how well it “fools” the discriminator. The generator is typically implemented via a latent variable z passed through a neural network G (a pushforward distribution), which turns the game into the common minG maxD formulation. The topic also emphasizes that, because the strategy spaces are not finite, equilibrium concepts can depend on move order (generator-first vs. discriminator-first), and the existence/uniqueness of equilibria becomes more subtle than in finite games. A key mathematical result in this framing is that for a fixed generator distribution μG, the optimal discriminator can be expressed using a Radon–Nikodym derivative, and the resulting value of the objective relates directly to the Jensen–Shannon divergence between the generator distribution and the reference (real) distribution. This connects the adversarial zero-sum game to a well-known statistical distance between distributions, clarifying what it means for the generator to match the data distribution.
GANs can be defined as a two-player zero-sum minimax game between a generator (minimizer) and a discriminator (maximizer).
The original GAN objective can be written in terms of expectations over the real distribution and the generator distribution, with discriminator outputs interpreted as probabilities of being real.
Equilibrium and move order are delicate in GANs because the strategy sets are not finitely spanned, so standard minimax interchange may fail.
For a fixed generator, the optimal discriminator has a closed-form expression (via Radon–Nikodym derivatives) and the game value relates to the Jensen–Shannon divergence between distributions.
A formulation where the generator’s objective is the discriminator’s loss, so one player’s improvement directly harms the other’s payoff.
A probability measure over the data space representing the distribution of samples produced by the generator.
A function or Markov kernel that maps inputs to probabilities in [0,1], indicating how likely an input is to be real.
The GAN loss defined so the generator minimizes L and the discriminator maximizes L.
A measure-theoretic tool used to express the optimal discriminator in terms of densities relative to a base measure.
A symmetric divergence between the real distribution and the generator distribution that appears in the optimal-discriminator value of the GAN objective.
The generator distribution induced by sampling latent variables z from μZ and transforming them through G.
“Can you explain what "GANs can be defined as a two-player zero-sum minimax game between a generator (minimizer) and a discriminator (maximizer)." means in simple terms?”