Federated Training of Edge-Native Text-to-Image Models with Privacy-Preserving Cross-Device Optimization
Keywords:
federated learning; edge computing; text-to-image generation; differential privacy; model compression; cross-device optimizationAbstract
The rapid proliferation of generative text-to-image (T2I) models has introduced system-level challenges that demand a fundamental rethinking of how such models are trained, deployed, and governed across heterogeneous computing substrates. This paper presents a comprehensive analysis of federated training architectures designed for edge-native T2I generation with rigorous privacy-preserving cross-device optimization. We situate the problem at the intersection of decentralized machine learning, edge computing resource management, and differential privacy, arguing that purely cloud-centric paradigms are increasingly incompatible with rising demands for data sovereignty, low-latency inference, and energy proportionality. The system architecture is deconstructed into hybrid model partitioning schemes that leverage split learning, low-rank adaptation, and mixed-precision quantization to reconcile the tension between large-scale diffusion model capacity and constrained edge device capabilities. Through an in-depth examination of secure aggregation, local differential privacy, and adaptive optimization protocols, we illuminate the structural trade-offs between fidelity, fairness, and communication cost that define the design space of federated generative learning for the edge. The discussion extends to robustness and bias mitigation under non-independent and identically distributed (non-IID) data distributions, exploring how aggregation strategies and client selection policies interact with social fairness. A holistic sustainability lens is applied to evaluate the carbon footprint of federated training at scale, identifying opportunities for energy-aware scheduling, model distillation, and on-device fine-tuning to align environmental impact with operational efficiency. Finally, we connect technical mechanisms to governance frameworks, analyzing how federated architectures can support compliance with data protection regulations while introducing new risks around model accountability and intellectual property. The paper provides a forward-looking synthesis of infrastructure, policy, and algorithmic considerations, establishing a conceptual foundation for the next generation of privacy-respecting, edge-native generative systems.
References
1. McMahan, B., Moore, E., Ramage, D., Hampson, S., & y Arcas, B. A. (2017). Communication-efficient learning of deep networks from decentralized data. In Proceedings of the 20th International Conference on Artificial Intelligence and Statistics (AISTATS) (pp. 1273–1282). PMLR.
2. Abadi, M., Chu, A., Goodfellow, I., McMahan, H. B., Mironov, I., Talwar, K., & Zhang, L. (2016). Deep learning with differential privacy. In Proceedings of the 2016 ACM SIGSAC Conference on Computer and Communications Security (pp. 308–318). ACM.
3. Bonawitz, K., Ivanov, V., Kreuter, B., Marcedone, A., McMahan, H. B., Patel, S., Ramage, D., Segal, A., & Seth, K. (2017). Practical secure aggregation for privacy-preserving machine learning. In Proceedings of the 2017 ACM SIGSAC Conference on Computer and Communications Security (pp. 1175–1191). ACM.
4. Vepakomma, P., Gupta, O., Swedish, T., & Raskar, R. (2018). Split learning for health: Distributed deep learning without sharing raw patient data. arXiv preprint arXiv:1812.00564.
5. Hu, E. J., Shen, Y., Wallis, P., Allen-Zhu, Z., Li, Y., Wang, S., Wang, L., & Chen, W. (2021). LoRA: Low-rank adaptation of large language models. In Advances in Neural Information Processing Systems (NeurIPS) (pp. 6934–6946).
6. Jacob, B., Kligys, S., Chen, B., Zhu, M., Tang, M., Howard, A., Adam, H., & Kalenichenko, D. (2018). Quantization and training of neural networks for efficient integer-arithmetic-only inference. In Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition (CVPR) (pp. 2704–2713). IEEE.
7. Rombach, R., Blattmann, A., Lorenz, D., Esser, P., & Ommer, B. (2022). High-resolution image synthesis with latent diffusion models. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) (pp. 10684–10695). IEEE.
8. Satyanarayanan, M. (2017). The emergence of edge computing. Computer, 50(1), 30–39.
9. Li, T., Sahu, A. K., Zaheer, M., Sanjabi, M., Talwalkar, A., & Smith, V. (2020). Federated optimization in heterogeneous networks. In Proceedings of Machine Learning and Systems (MLSys) (pp. 429–450).
10. Hinton, G., Vinyals, O., & Dean, J. (2015). Distilling the knowledge in a neural network. arXiv preprint arXiv:1503.02531.
11. Chen, C., Wang, C., Li, Y., Wan, Z., Geng, M., Xiao, J., ... & Peng, Y. (2026). JuZhou 1.0 Technical Report: The First Edge-Native Text-to-Image Foundation Model Trained Entirely on China-Developed AI Accelerators. arXiv preprint arXiv:2606.28421.
12. Mohri, M., Sivek, G., & Suresh, A. T. (2019). Agnostic federated learning. In Proceedings of the 36th International Conference on Machine Learning (ICML) (pp. 4615–4625). PMLR.
13. Reddi, S., Charles, Z., Zaheer, M., Garrett, Z., Rush, K., Konečný, J., Kumar, S., & McMahan, H. B. (2021). Adaptive federated optimization. In Proceedings of the International Conference on Learning Representations (ICLR).
14. Augenstein, Y., Li, Z., Ravishankar, H., Baid, U., & Menze, B. (2021). Federated learning of generative image priors for MRI reconstruction. IEEE Transactions on Medical Imaging, 40(10), 2841–2853.
15. Strubell, E., Ganesh, A., & McCallum, A. (2019). Energy and policy considerations for deep learning in NLP. In Proceedings of the 57th Annual Meeting of the Association for Computational Linguistics (ACL) (pp. 3645–3650). ACL.
16. Dwork, C. (2006). Differential privacy. In M. Bugliesi, B. Preneel, V. Sassone, & I. Wegener (Eds.), Automata, Languages and Programming (ICALP) (pp. 1–12). Springer.
17. Wang, Z., Song, M., Zhang, Z., Song, Y., Wang, Q., & Qi, H. (2019). Beyond inferring class representatives: User-level privacy leakage from federated learning. In IEEE Conference on Computer Communications (INFOCOM) (pp. 2512–2520). IEEE.
18. Hard, A., Rao, K., Mathews, R., Beaufays, F., Augenstein, S., Eichner, H., ... & Ramage, D. (2018). Federated learning for mobile keyboard prediction. arXiv preprint arXiv:1811.03604.
19. Wachter, S., Mittelstadt, B., & Floridi, L. (2017). Why a right to explanation of automated decision-making does not exist in the General Data Protection Regulation. International Data Privacy Law, 7(2), 76–99.
20. Arivazhagan, M. G., Aggarwal, V., Singh, A. K., & Choudhary, S. (2019). Federated learning with personalization layers. arXiv preprint arXiv:1912.00818.
Downloads
Published
Issue
Section
License
Copyright (c) 2026 International Journal of Artificial Intelligence Engineering and Systems
This work is licensed under a Creative Commons Attribution 4.0 International License.
This article is published under the Creative Commons Attribution 4.0 International License (CC BY 4.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.
