PP-FCIL: A Dual-Backbone Privacy-Preserving Federated Class-Incremental Learning Framework with Adaptive Feature Fusion
International Journal of Innovative Science and Research Technology · 30.04.2026 · 2566
DOI: 10.38124/ijisrt/26apr1499 ↗Kurz gefasst
Fachartikel in der Zeitschrift International Journal of Innovative Science and Research Technology, online erschienen am 30.04.2026. Verfasst von Darsi Venkata Varalakshmi und 3 weiteren Autor:innen. Die Arbeit nennt 15 Literaturangaben.
Abstract (Originalsprache)
Federated learning provides an option to train models cooperatively without the need for sharing data centrally. Yet, incorporating class-incremental learning in federated learning gives rise to challenges like catastrophic forgetting, heterogeneity of the data, and data privacy concerns. To solve these problems, this paper presents a new method of Privacy-Preserving Federated Class-Incremental Learning (PP-FCIL), where the method is tested against the CIFAR-100 benchmark. The proposed solution uses a two-head backbone and channel attention mechanism to retain prior knowledge as well as adapt to newly introduced classes in each stage of learning. Also, to prevent forgetting and class imbalance, this paper proposes exemplar memory using herding, knowledge distillation, supervised contrastive learning, and balanced softmax loss as part of one training process. Different from traditional federated learning methods, the proposed framework utilizes Bayesian Differential Privacy in the clients that clips gradients and adds Gaussian noise to protect data privacy. Furthermore, to measure cumulative privacy loss, a strict privacy accounting protocol is used. Also, a multifactor aggregation protocol is adopted to weigh the importance of client contribution in a non-IID setting.
Zitieren
APA
Varalakshmi, D. V., Sangeetha, G., Nikhitha, D., & Bai, Z. S. (2026). PP-FCIL: A Dual-Backbone Privacy-Preserving Federated Class-Incremental Learning Framework with Adaptive Feature Fusion. International Journal of Innovative Science and Research Technology, 2566. https://doi.org/10.38124/ijisrt/26apr1499
Vancouver
Varalakshmi DV, Sangeetha G, Nikhitha D, Bai ZS. PP-FCIL: A Dual-Backbone Privacy-Preserving Federated Class-Incremental Learning Framework with Adaptive Feature Fusion. International Journal of Innovative Science and Research Technology. 2026:2566. doi:10.38124/ijisrt/26apr1499
BibTeX
@article{Varalakshmi2026,
author = {Varalakshmi, Darsi Venkata and Sangeetha, Garige and Nikhitha, Dammu and Bai, Z. Sunitha},
title = {PP-FCIL: A Dual-Backbone Privacy-Preserving Federated Class-Incremental Learning Framework with Adaptive Feature Fusion},
journal = {International Journal of Innovative Science and Research Technology},
year = {2026},
pages = {2566},
doi = {10.38124/ijisrt/26apr1499}
}
EUVIMED