Generalization of GANs and overparameterized models under Lipschitz continuity

no code implementations • 6 Apr 2021 • Khoat Than, Nghia Vu

This result answers the long mystery of why the popular use of Lipschitz constraint for GANs often leads to great success, empirically without a solid theory.

Data Augmentation Generalization Bounds

Generalization and Stability of GANs: A theory and promise from data augmentation

no code implementations • 1 Jan 2021 • Khoat Than, Nghia Vu

Finally, we show why data augmentation can ensure Lipschitz continuity on both the discriminator and generator.

Data Augmentation