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.
Related benchmarks
| Task | Dataset | Result | Rank | |
|---|---|---|---|---|
| Image Classification | CIFAR-10 IID | Accuracy50.58 | 185 | |
| Image Classification | CIFAR-10 Dir(0.5) | Accuracy38.95 | 59 | |
| Image Classification | Fashion MNIST | Gini Coefficient0.3966 | 42 | |
| Image Classification | MNIST-1797 | Gini Coefficient0.0132 | 42 | |
| Image Classification | CIFAR-10 Dir-0.1 | Accuracy25.93 | 34 | |
| Image Classification | MNIST (Dir(0.5)) | Accuracy86.63 | 19 | |
| Image Classification | FashionMNIST (IID) | Accuracy87.88 | 17 | |
| Image Classification | MNIST 1797 LS K=1 | Top-1 Accuracy96.95 | 6 | |
| Federated Learning | CIFAR-10 IID | Gini Coefficient0.0153 | 6 | |
| Federated Learning | CIFAR-10 Dir 0.8 | Gini Coefficient6.84 | 6 |