ML: Gan

Machine-learning ideas tagged Gan in the ML taxonomy of the Math2NN corpus.

2 ideas found

ML: Gan Math: Stochastic processes Clear all

△ Mechanism confirmed, baseline not beaten 2026

Tikhonov-Stabilized Stochastic Extragradient

Replace the raw stochastic saddle objective by a strongly convex-strongly concave, quadratically anchored objective before applying stochastic extragradient. For a generator-discriminator or policy-rewarder game, anchor the minimizing and maximizing parameter vectors to reference parameters with opposite signs, suppressing persistent stochastic rotations and improving the quality of the final iterate.

Useful7/10

Difficulty4/10

Novelty6/10

Optimization Stability Accuracy Sample efficiency

Paper: Last-Iterate Convergence of Single-Loop Stochastic Methods for Constrained Convex-Concave Minimax Problems arXiv:2607.11056

Unverified 2026

Accumulated Tikhonov Extragradient for Stochastic Games

Treat a two-player neural game as a stochastic monotone operator problem and add an anchor regularizer that makes the current game strongly monotone during early training. Use stochastic extragradient steps inside geometrically decreasing regularization stages, retaining the accumulated anchor rather than resetting it; this should suppress rotational GAN dynamics while avoiding a permanent bias toward the initialization.

Useful6/10

Difficulty5/10

Novelty6/10

Optimization Stability Speedup Generalization

Paper: Computation of Strong Solutions to Stochastic Variational Inequalities arXiv:2609.04188