Communication-Efficient and Robust Multi-Modal Federated Learning via Latent-Space Consensus
IEEE Wireless Communications Letters · 2026 · 15 · 2298-2302
DOI: 10.1109/lwc.2026.3676885 ↗Kurz gefasst
Fachartikel in der Zeitschrift IEEE Wireless Communications Letters, 2026. Verfasst von Mohamed Badi und 2 weiteren Autor:innen. Die Arbeit nennt 11 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0). Gefördert von: ERA-NET CHIST-ERA under Project MUSE-COM².
Abstract (Originalsprache)
Federated learning (FL) enables collaborative model training across distributed devices without sharing raw data, but applying FL to multi-modal settings introduces significant challenges. Clients typically possess heterogeneous modalities and model architectures, making it difficult to align feature spaces efficiently while preserving privacy and minimizing communication costs. To address this, we introduce CoMFed, a Communication-Efficient Multi-Modal Federated Learning framework that uses learnable projection matrices to generate compressed latent representations. A latent-space regularizer aligns these representations across clients, improving cross-modal consistency and robustness to outliers. Experiments on human activity recognition benchmarks show that CoMFed achieves competitive accuracy with minimal overhead.
Förderung
- ERA-NET CHIST-ERA under Project MUSE-COM²
Zitieren
APA
Badi, M., Issaid, C. B., & Bennis, M. (2026). Communication-Efficient and Robust Multi-Modal Federated Learning via Latent-Space Consensus. IEEE Wireless Communications Letters, 15, 2298-2302. https://doi.org/10.1109/lwc.2026.3676885
Vancouver
Badi M, Issaid CB, Bennis M. Communication-Efficient and Robust Multi-Modal Federated Learning via Latent-Space Consensus. IEEE Wireless Communications Letters. 2026;15:2298-2302. doi:10.1109/lwc.2026.3676885
BibTeX
@article{Badi2026,
author = {Badi, Mohamed and Issaid, Chaouki Ben and Bennis, Mehdi},
title = {Communication-Efficient and Robust Multi-Modal Federated Learning via Latent-Space Consensus},
journal = {IEEE Wireless Communications Letters},
year = {2026},
volume = {15},
pages = {2298--2302},
doi = {10.1109/lwc.2026.3676885}
}
EUVIMED