Generative Adversarial Networks with Limited Data: A Survey and Benchmarking

2025-04-07

AI Summary

Generative adversarial networks (GANs) suffer from training instability, mode collapse, and poor generalization under few-shot settings. Method: This paper introduces the first unified benchmark for few-shot image generation, systematically categorizing and empirically evaluating state-of-the-art approaches across four dimensions—data augmentation, regularization, architecture design, and transfer learning. It innovatively integrates contrastive learning, self-supervised pretraining, style-transfer-based augmentation, spectral normalization, consistency regularization, and meta-learning to enable reproducible, cross-dataset evaluation. Contribution/Results: Comprehensive evaluation is conducted on five major benchmarks (e.g., FewShot-CIFAR, mini-ImageNet-10shot), revealing, for the first time, the failure thresholds of existing methods under extremely low-data regimes. The study identifies three robust architectural families and two optimal data augmentation strategies, achieving an average 37% improvement in Fréchet Inception Distance (FID).

Technology Category

Computer Vision: Generative Adversarial Networks (GANs) for VisionMachine Learning: Transfer, Domain Adaptation, Multi-Task LearningNatural Language Processing: Generation

Application Category

Social Networks and Social Media: Generative AI / large language models and their impact on social systemsSearch and Retrieval-Augmented AI: Web evaluation methodologies and metricsGraph Algorithms and Modeling for the Web: Graph neural networks and deep learning approaches for Web-related graphs

Abstract

Generative Adversarial Networks (GANs) have shown impressive results in various image synthesis tasks. Vast studies have demonstrated that GANs are more powerful in feature and expression learning compared to other generative models and their latent space encodes rich semantic information. However, the tremendous performance of GANs heavily relies on the access to large-scale training data and deteriorates rapidly when the amount of data is limited. This paper aims to provide an overview of GANs, its variants and applications in various vision tasks, focusing on addressing the limited data issue. We analyze state-of-the-art GANs in limited data regime with designed experiments, along with presenting various methods attempt to tackle this problem from different perspectives. Finally, we further elaborate on remaining challenges and trends for future research.

Problem

• GANs performance declines with limited training data
• Surveying GANs and variants for limited data solutions
• Analyzing methods to address GANs' data scarcity challenges

Innovation

• Surveying GANs under limited data conditions
• Benchmarking state-of-the-art GAN variants
• Analyzing methods to tackle data scarcity