Share your thoughts, 1 month free Claude Pro on usSee more
WorkDL logo mark

FIRMA: FIbonacci Ring Model Aggregation for Privacy-preserving Federated Learning

About

Federated learning protocols face a structural trilemma: canonical server-based aggregation~\cite{mcmahan2017} creates a single point of failure and gradient inversion risk; decentralised ring-gossip alternatives~\cite{hu2019segmented} expose classification heads to semi-honest peers via uninformed uniform weights; and personalised methods~\cite{collins2021exploiting} reintroduce central aggregation. No existing protocol simultaneously achieves server-free operation, permanently private heads, ring topology, and principled asymmetric neighbour weighting. We propose FIRMA (\textbf{FI}bonacci \textbf{R}ing \textbf{M}odel \textbf{A}ggregation), a family of three progressively enhanced federated learning protocols: 1) \fibfl\ establishes the foundation: server-free ring aggregation with Fibonacci-weighted neighbour blending and permanently private classification heads. 2) \fibflp\ augments this with accuracy-gated neighbour suppression, selectively down-weighting poorly-converged peers while preserving the Fibonacci directional bias. 3) \fibflpp, the full system, completes the family with a 2-opt ring permutation that maximises adjacent-client class diversity, global ring coverage via $K_g{=}\lceil N/2\rceil$ gossip passes, and cosine-annealed self-retention calibration. We establish a convergence rate bound and three supporting propositions governing normalisation, coverage, retention, and diversity optimality. Systematic experiments across 28 configurations -- four benchmarks crossed with seven heterogeneity regimes -- demonstrate that \fibflpp\ surpasses \fedavg\ in all 12 label-skew configurations, with a peak advantage of $+20.7$\,pp on CIFAR-10 at $K{=}1$. Under Dirichlet heterogeneity, \fibflpp\ is the Pareto-dominant method among all server-free protocols, achieving the highest accuracy in 17 of 28 configurations.

Rachid Hedjam• 2026

Related benchmarks

TaskDatasetResultRank
Image ClassificationCIFAR-10 IID
Accuracy50.58
185
Image ClassificationCIFAR-10 Dir(0.5)
Accuracy38.95
59
Image ClassificationFashion MNIST
Gini Coefficient0.3966
42
Image ClassificationMNIST-1797
Gini Coefficient0.0132
42
Image ClassificationCIFAR-10 Dir-0.1
Accuracy25.93
34
Image ClassificationMNIST (Dir(0.5))
Accuracy86.63
19
Image ClassificationFashionMNIST (IID)
Accuracy87.88
17
Image ClassificationMNIST 1797 LS K=1
Top-1 Accuracy96.95
6
Federated LearningCIFAR-10 IID
Gini Coefficient0.0153
6
Federated LearningCIFAR-10 Dir 0.8
Gini Coefficient6.84
6
Showing 10 of 44 rows

Other info

Follow for update