FSCO multiplies the discriminator learning rate by a DDPG-chosen factor in [0,1], using reward -|G_loss - D_loss|, and reports more stable-looking GAN training on three image datasets.
Title resolution pending
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.LG 1years
2025 1verdicts
REJECT 1representative citing papers
citing papers explorer
-
Fast-Slow Co-advancing Optimizer: Toward Harmonious Adversarial Training of GAN
FSCO multiplies the discriminator learning rate by a DDPG-chosen factor in [0,1], using reward -|G_loss - D_loss|, and reports more stable-looking GAN training on three image datasets.