Diverse Rare Sample Generation with Pretrained GANs
This addresses the issue of limited diversity and coverage in generative models for applications requiring rare data, though it is incremental as it builds on existing GAN frameworks.
The study tackled the problem of generating diverse rare samples from high-resolution image datasets using pretrained GANs, achieving controllable generation of rare images without retraining the GANs.
Deep generative models are proficient in generating realistic data but struggle with producing rare samples in low density regions due to their scarcity of training datasets and the mode collapse problem. While recent methods aim to improve the fidelity of generated samples, they often reduce diversity and coverage by ignoring rare and novel samples. This study proposes a novel approach for generating diverse rare samples from high-resolution image datasets with pretrained GANs. Our method employs gradient-based optimization of latent vectors within a multi-objective framework and utilizes normalizing flows for density estimation on the feature space. This enables the generation of diverse rare images, with controllable parameters for rarity, diversity, and similarity to a reference image. We demonstrate the effectiveness of our approach both qualitatively and quantitatively across various datasets and GANs without retraining or fine-tuning the pretrained GANs.