IRO Journals
Journal of Innovative Image Processing

Diversity-Preserving Few-Shot GAN Adaptation via Layer-Aware and Similarity-Guided Meta-Training Samples

Open Access
Published: 18 July, 2026
Pages: 1096-1115
  • Shoeb Ahmad , Shoeb Ahmad
    Research Scholar, Department of Computer Science and Engineering, JAIN (Deemed-to-be University), Kanakapura Campus, Bangalore, Karnataka, India
    Research Scholar, Department of Computer Science and Engineering, JAIN (Deemed-to-be University), Kanakapura Campus, Bangalore, Karnataka, India
  • Gowrishankar J. , Gowrishankar J.
    Department of Computer Science and Engineering, Faculty of Engineering and Technology, JAIN (Deemed-to-be University), Kanakapura Campus, Bangalore, Karnataka, India
    Department of Computer Science and Engineering, Faculty of Engineering and Technology, JAIN (Deemed-to-be University), Kanakapura Campus, Bangalore, Karnataka, India
  • Sonal Sharma Sonal Sharma
    Department of Computer Science and Engineering, Faculty of Engineering and Technology, JAIN (Deemed-to-be University), Kanakapura Campus, Bangalore, Karnataka, India
    Department of Computer Science and Engineering, Faculty of Engineering and Technology, JAIN (Deemed-to-be University), Kanakapura Campus, Bangalore, Karnataka, India
view PDF
view PDF

How to Cite

Ahmad, Shoeb, Gowrishankar J., and Sonal Sharma. 2026. “Diversity-Preserving Few-Shot GAN Adaptation via Layer-Aware and Similarity-Guided Meta-Training Samples”. Journal of Innovative Image Processing 8 (3): 1096-1115. https://doi.org/10.36548/jiip.2026.3.017.
ACM ACS APA ABNT Chicago Harvard IEEE MLA Turabian Vancouver AMA

Download Citation

Endnote/Zotero/Mendeley (RIS) BibTeX

Keywords

Meta-Learning
Reptile-MetaGAN
Generative Adversarial Networks (GANs)
Few-shot Learning
First-Order Meta-Learning
Layer-Wise Fine-Tuning

Abstract

Few-shot learning is a significant challenge for Meta-GANs. Without explicit mechanisms to preserve diversity, the model tends to overfit to the scarce training data. We build on a Reptile-based meta-learning backbone and extend it into a diversity-preserving Meta-GAN framework. Our model aims to improve both the stability and diversity of few-shot adaptation. Although the meta-training phase follows the standard Reptile mechanism, the meta-testing phase introduces several key modifications to address the diversity collapse that is typical of low-data adaptation. We employ a strategy in which one copy of the meta-trained model is reused to induce diversity throughout the meta-test training, whereas another copy is adapted to the target task. To ensure fidelity to the target sample, we use layer freezing and train exclusively on the target sample and meta-training samples during alternating discriminator iterations. Compared to baseline Reptile-MetaGAN experiments on the MNIST and Omniglot datasets in a few-shot configuration yielded improvements in Fréchet Inception Distance (FID), Kernel Inception Distance (KID), and Inception Score (IS). These findings demonstrate that a stable and diversity-aware pathway for few-shot generative adaptation can be achieved using cross-task diversity meta-train samples, alternating layer freezing, and selective SSIM filtering.

References

  1. Nichol, Alex, Joshua Achiam, and John Schulman. "On First-Order Meta-Learning Algorithms." arXiv preprint arXiv:1803.02999 (2018).
  2. Clouâtre, Louis, and Marc Demers. "Figr: Few-Shot Image Generation with Reptile." arXiv preprint arXiv:1901.02199 (2019).
  3. Mo, Sangwoo, Minsu Cho, and Jinwoo Shin. "Freeze the Discriminator: A Simple Baseline for Fine-Tuning Gans." arXiv preprint arXiv:2002.10964 (2020).
  4. Zhang, Ruixiang, Tong Che, Zoubin Ghahramani, Yoshua Bengio, and Yangqiu Song. "Metagan: An Adversarial Approach to Few-Shot Learning." Advances in neural information processing systems 31 (2018).
  5. Goodfellow, Ian J., Jean Pouget-Abadie, Mehdi Mirza, Bing Xu, David Warde-Farley, Sherjil Ozair, Aaron Courville, and Yoshua Bengio. "Generative Adversarial Nets." Advances in neural information processing systems 27 (2014).
  6. Gulrajani, Ishaan, Faruk Ahmed, Martin Arjovsky, Vincent Dumoulin, and Aaron C. Courville. "Improved Training of Wasserstein Gans." Advances in neural information processing systems 30 (2017).
  7. Yang, Mengping, and Zhe Wang. "Image Synthesis Under Limited Data: A Survey and Taxonomy." International Journal of Computer Vision 133, no. 6 (2025): 3689-3726.
  8. Mikolaj Binkowski, Danica J. Sutherland, Michael Arbel and Arthur Gretton “Demystifying MMD GANs.” 6th International Conference on Learning Representations, ICLR, 2018.
  9. Heusel, Martin, Hubert Ramsauer, Thomas Unterthiner, Bernhard Nessler, and Sepp Hochreiter. "Gans Trained by A Two Time-Scale Update Rule Converge to a Local Nash Equilibrium." Advances in neural information processing systems 30 (2017).
  10. Chavdarova, Tatjana, Matteo Pagliardini, Sebastian U. Stich, François Fleuret, and Martin Jaggi. "Taming Gans with Lookahead-Minmax." arXiv preprint arXiv:2006.14567 (2020).
  11. Lin, Alex Tong, Wuchen Li, Stanley Osher, and Guido Montúfar. "Wasserstein Proximal of Gans." In International Conference on Geometric Science of Information, pp. 524-533. Cham: Springer International Publishing, 2021.
  12. Ojha, Utkarsh, Yijun Li, Jingwan Lu, Alexei A. Efros, Yong Jae Lee, Eli Shechtman, and Richard Zhang. "Few-Shot Image Generation via Cross-Domain Correspondence." In Proceedings of the IEEE/CVF conference on computer vision and pattern recognition, 2021, 10743-10752.
  13. Giannone, Giorgio, Didrik Nielsen, and Ole Winther. "Few-Shot Diffusion Models." arXiv preprint arXiv:2205.15463 (2022).
  14. Bartunov, Sergey, and Dmitry Vetrov. "Few-Shot Generative Modelling with Generative Matching Networks." In International conference on artificial intelligence and statistics, PMLR, 2018, 670-678.
  15. Karras, Tero, Miika Aittala, Janne Hellsten, Samuli Laine, Jaakko Lehtinen, and Timo Aila. "Training Generative Adversarial Networks with Limited Data." Advances in neural information processing systems 33 (2020): 12104-12114.
  16. Miyato, Takeru, Toshiki Kataoka, Masanori Koyama, and Yuichi Yoshida. "Spectral Normalization for Generative Adversarial Networks." arXiv preprint arXiv:1802.05957 (2018).
  17. Salimans, Tim, Ian Goodfellow, Wojciech Zaremba, Vicki Cheung, Alec Radford, and Xi Chen. "Improved Techniques for Training Gans." Advances in neural information processing systems 29 (2016).
  18. Hospedales, Timothy, Antreas Antoniou, Paul Micaelli, and Amos Storkey. "Meta-Learning in Neural Networks: A Survey." IEEE transactions on pattern analysis and machine intelligence 44, no. 9 (2021): 5149-5169.
  19. Wang, Yaxing, Chenshen Wu, Luis Herranz, Joost Van de Weijer, Abel Gonzalez-Garcia, and Bogdan Raducanu. "Transferring Gans: Generating Images from Limited Data." In Proceedings of the European conference on computer vision (ECCV), 2018, 218-234.
  20. Wang, Yaxing, Abel Gonzalez-Garcia, David Berga, Luis Herranz, Fahad Shahbaz Khan, and Joost van de Weijer. "Minegan: Effective Knowledge Transfer from Gans to Target Domains with Few Images." In Proceedings of the IEEE/CVF conference on computer vision and pattern recognition, 2020, 9332-9341.
  21. hou Wang, A. C. Bovik, H. R. Sheikh and E. P. Simoncelli, "Image Quality Assessment: from Error Visibility to Structural Similarity," in IEEE Transactions on Image Processing, vol. 13, no. 4, April 2004, 600-612.
  22. Zhang, Michael, James Lucas, Jimmy Ba, and Geoffrey E. Hinton. "Lookahead Optimizer: K Steps Forward, 1 Step Back." Advances in neural information processing systems 32 (2019).
  23. Zhang, Yabo, Yuxiang Wei, Zhilong Ji, Jinfeng Bai, and Wangmeng Zuo. "Towards Diverse and Faithful One-Shot Adaption of Generative Adversarial Networks." Advances in Neural Information Processing Systems 35 (2022): 37297-37308.
  24. LeCun, Yann, Léon Bottou, Yoshua Bengio, and Patrick Haffner. "Gradient-Based Learning Applied to Document Recognition." Proceedings of the IEEE 86, no. 11 (1998): 2278-2324.
  25. Lake, Brenden M., Ruslan Salakhutdinov, and Joshua B. Tenenbaum. "Human-Level Concept Learning Through Probabilistic Program Induction." Science 350, no. 6266 (2015): 1332-1338.