Copyright © 2026 Authors retain the copyright of this article. This article is an open access article distributed under the Creative Commons Attribution License which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.

@article{207420,
        author = {Sujay S and Kavyashree H and Kavya T N},
        title = {Explainable Privacy-Preserving Federated Learning Using Differential Privacy and FedProx for Non-IID Image Classification},
        journal = {International Journal of Innovative Research in Technology},
        year = {2026},
        volume = {13},
        number = {3},
        pages = {773-780},
        issn = {2349-6002},
        url = {https://ijirt.org/article?manuscript=207420},
        abstract = {Federated Learning (FL) is well suited for privacy-sensitive applications because it enables collaborative model training without requiring clients to share raw data. How-ever, real-world federated environments often contain Non-Independent and Identically Distributed (Non-IID) data, which can lead to optimization instability, privacy concerns, and limited model interpretability. This paper presents an explain-able privacy-preserving federated learning framework for image classification by integrating Differential Privacy (DP), FedProx optimization, and GradCAM-based explainability. The proposed framework is evaluated on CIFAR-10 and CIFAR-100 datasets under different federated configurations, including FedAvg, Fed-Prox, DP-FedAvg, and DP-FedProx.
The experimental results show that FedProx improves op-timization robustness in non-IID settings, while Differential Privacy introduces a measurable trade-off between privacy pro-taction and predictive utility. The explainability analysis further indicates that privacy-preserving optimization produces more spatially distributed activation patterns compared with non-private models. These findings highlight the close relationship between optimization stability, privacy preservation, and inter-pretability, supporting the development of more transparent and trustworthy federated learning systems.},
        keywords = {Federated Learning, Differential Privacy, Fed-Prox, Explainable Artificial Intelligence, GradCAM, Non-IID Learning, Image Classification, Privacy-Preserving Machine Learning.},
        month = {August},
        }