“Polarity Sampling: Quality and Diversity Control of Pre-Trained Generative Networks via Singular Values”, Ahmed Imtiaz Humayun, Randall Balestriero, Richard Baraniuk2022-03-03 (BigGAN, StyleGAN, autoencoder NN; similar)⁠:

We present Polarity Sampling, a theoretically justified plug-and-play method for controlling the generation quality and diversity of pre-trained deep generative networks (DGNs).

Leveraging the fact that DGNs are, or can be approximated by, continuous piecewise affine splines, we derive the analytical DGN output space distribution as a function of the product of the DGN’s Jacobian singular values raised to a power ρ. We dub ρ the polarity parameter and prove that ρ focuses the DGN sampling on the modes (ρ < 0) or anti-modes (ρ > 0) of the DGN output-space distribution.

We demonstrate that nonzero polarity values achieve a better precision-recall (quality-diversity) Pareto frontier than standard methods, such as truncation, for a number of state-of-the-art DGNs. We also present quantitative and qualitative results on the improvement of overall generation quality (eg. in terms of the Fréchet Inception Distance) for a number of state-of-the-art DGNs, including StyleGAN3, BigGAN-deep, NVAE, for different conditional and unconditional image generation tasks. In particular, Polarity Sampling redefines the state-of-the-art for StyleGAN-2 on the FFHQ Dataset to FID 2.57, StyleGAN-2 on the LSUN Car Dataset to FID 2.27 and StyleGAN3 on the AFHQv2 Dataset to FID 3.95.

Demo link.

Similar Links: