arXiv：cs.LG· Alessandro Licciardi·· 3 小时前AI 评分35
RIPPLE：基于小波散射变换的联邦学习零样本聚类
RIPPLE in Still Water: Zero-Shot Clustering in Federated Learning with Wavelet Scattering Transform
AI 导读
RIPPLE 提出一种聚类联邦学习框架，聚类分配完全离线计算，基于小波散射变换嵌入的方差加权主成分原型，由服务端预训练的高斯混合 VAE 解码。其每轮通信开销与 FedAvg 完全一致，未参与训练的客户端仅需一次前向传播即可获得个性化模型。在五个基准上 RIPPLE 持续优于所有基线，该工作已被 NeurIPS 2026 主赛道接收。
正文
Abstract:Clustered Federated Learning (FL) partitions a client population into groups of similar local distributions and trains one specialized model per cluster, mitigating client drift that degrades single-model methods under non-IID data. Prior methods discover cluster structure inside the training loop through gradient similarity, loss evaluation, or EM-style updates, thus increasing communication overhead, exposing gradients to inversion attacks, and providing no mechanism to assign clients absent from training. We propose RIPPLE, a clustered FL framework in which cluster assignment is computed entirely offline from a spectral characterization of each client's local data: a variance-weighted principal-component prototype embedded via the Wavelet Scattering Transform and decoded by a Gaussian Mixture VAE trained server-side on synthetic client populations before federation begins. Per-round communication cost matches FedAvg exactly, and a client absent from training obtains a personalized model from a single forward pass, without gradient computation, model evaluation, or extra communication round. We prove that the gap between RIPPLE's surrogate clustered objective and the oracle is bounded by a computable quantity decaying with client sample size and independent of federation duration; per-cluster convergence matches the minimax-optimal rate for non-convex smooth objectives. Across five benchmarks spanning controlled and realistic heterogeneity, RIPPLE consistently outperforms all baselines, with margins growing on the most realistic partitions.
| Comments: | Accepted at NeurIPS 2026 (main track) |
| Subjects: | Machine Learning (cs.LG) |
| Cite as: | arXiv:2610.03054 [cs.LG] |
| (or arXiv:2610.03054v1 [cs.LG] for this version) | |
| https://doi.org/10.48550/arXiv.2610.03054 arXiv-issued DOI via DataCite (pending registration) |
Submission history
From: Alessandro Licciardi [view email]
[v1]
Fri, 2 Oct 2026 09:35:39 UTC (407 KB)
来源：arXiv：cs.LG · arxiv.org