Diverse Rare Sample Generation with Pretrained GANs

TL;DR

This paper introduces a novel method for generating diverse rare samples from high-resolution datasets using pretrained GANs, employing gradient-based optimization and normalizing flows to enhance diversity and control without retraining.

Contribution

It presents a new approach combining gradient optimization and normalizing flows to generate rare, diverse samples from pretrained GANs without additional training.

Findings

Effective generation of diverse rare images demonstrated

Method works across various datasets and GAN architectures

No retraining or fine-tuning required

Abstract

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.