MineGAN++: Mining Generative Models for Efficient Knowledge Transfer to Limited Data Domains
This addresses the challenge of applying GANs to small target datasets, which is incremental but practical for domains with limited data.
The authors tackled the problem of efficiently transferring knowledge from pretrained GANs to limited data domains by proposing MineGAN, which uses a miner network to identify beneficial generative distributions and sparse subnetwork selection to prevent overfitting, resulting in outperformance over existing methods on challenging datasets.
GANs largely increases the potential impact of generative models. Therefore, we propose a novel knowledge transfer method for generative models based on mining the knowledge that is most beneficial to a specific target domain, either from a single or multiple pretrained GANs. This is done using a miner network that identifies which part of the generative distribution of each pretrained GAN outputs samples closest to the target domain. Mining effectively steers GAN sampling towards suitable regions of the latent space, which facilitates the posterior finetuning and avoids pathologies of other methods, such as mode collapse and lack of flexibility. Furthermore, to prevent overfitting on small target domains, we introduce sparse subnetwork selection, that restricts the set of trainable neurons to those that are relevant for the target dataset. We perform comprehensive experiments on several challenging datasets using various GAN architectures (BigGAN, Progressive GAN, and StyleGAN) and show that the proposed method, called MineGAN, effectively transfers knowledge to domains with few target images, outperforming existing methods. In addition, MineGAN can successfully transfer knowledge from multiple pretrained GANs.