gan-complete-theoretical-characterization
IN derived (depth 1)
Created 2026-06-21T09:59:01+00:00 · Reviewed 2026-06-21T15:37:01+00:00
GAN theory provides a complete analytical characterization of optimal behavior: the optimal discriminator has a closed-form solution (Radon-Nikodym derivative), the objective implicitly minimizes Jensen-Shannon divergence, and at the unique equilibrium the generator exactly recovers the data distribution.
Justifications
SL — Three theoretical results fully characterize GAN optimality from discriminator to equilibrium
Antecedents (all must be IN):
- IN gan-optimal-discriminator-formula — The optimal discriminator for a fixed generator is D*(x) = μ_ref(x) / (μ_ref(x) + μ_G(x)), which is the Radon-Nikodym derivative dμ_ref/d(μ_ref + μ_G)
- IN gan-minimizes-jensen-shannon-divergence — The original GAN implicitly minimizes the Jensen-Shannon divergence (not KL divergence) between the generated and real data distributions
- IN gan-equilibrium-generator-matches-data — At the unique GAN equilibrium, the generator matches the data distribution exactly (μ̂_G = μ_ref), the discriminator outputs 1/2 everywhere, and the objective value is −2 ln 2
Dependents
These beliefs depend on this one:
- OUT gan-theory-practice-alignment — GAN theory and practice are aligned — the complete analytical characterization (optimal discriminator, JSD minimization, unique equilibrium) accurately predicts GAN behavior in real-world deployment across physics simulation, medical imaging, and creative applications.
- OUT gan-training-stabilizable-given-complete-theory — GAN training would be reliably stabilizable given the complete theoretical characterization (optimal discriminator, JSD minimization, unique equilibrium) and multiple complementary stabilization interventions (non-saturating loss, TTUR, architecture choices) — unless the game-theoretic foundations themselves are limited, with Nash equilibria not guaranteed in general and equilibrium equivalence holding only for the original formulation.
- IN theoretical-completeness-no-guarantee-of-paradigm-durability — Theoretical completeness does not guarantee paradigm durability — GANs had a notably complete analytical characterization (closed-form optimal discriminator, JSD minimization proof, unique equilibrium) yet were largely supplanted by diffusion models from approximately 2022 onward, suggesting that factors beyond theoretical elegance — possibly including training reliability — may play a significant role in determining which paradigms persist.