Federated clustering finds unknown group counts on local and global data
Federated Clustering with Unknown Local and Global Cluster Cardinalities
Machine LearningDistributed, Parallel, and Cluster Computing
Summary
Clustering groups similar data points, but in federated settings, neither the central server nor the individual clients know how many groups should exist locally or globally. The authors propose a two-step method where clients first guess their local number of groups from their data, and then a server combines these guesses to find global clusters. They introduce a new technique called Adaptive Split–Merge (ASM) that grows and adjusts group counts without needing labeled data or knowing the true number beforehand. This approach works well across diverse datasets and helps in situations where clients have very different data sizes and unknown group counts.
What this means in practice
- •For industrial data engineers: Identify faulty machine groups across independently operated factories without knowing how many fault types appear locally or globally.
- •For distributed systems engineers: Combine clustering results from multiple independent clients with unknown cluster counts to form reliable global groups while preserving privacy.
Authors
Mitushi Goyal, Tarun S., Riddhanya Senapathi, Arun Raman
Abstract
Federated clustering methods that do not require the global number of clusters $K$ still assume that each client knows its local number $K_g$. This assumption is hard to justify when clients know no more about their data than the server does, as in fault diagnosis across independently operated industrial sites. We propose a two-phase framework in which neither count is known: each client first estimates $K_g$ from its own data, and an aggregator that requires local counts, such as FedGEM, then uses these estimates in place of the true values. For the first phase we introduce Adaptive Split--Merge (ASM), which grows a spherical Gaussian mixture by BIC-driven splitting and then merges excess components. ASM uses no labels, selects its hyperparameters on held-out client data only, and makes no assumption about how clusters are shared across clients. We derive a closed-form split criterion whose critical cluster size falls with anisotropy and rises with dimension, and show empirically that over-fragmentation grows with the number of points per cluster, which federation divides among clients. Across eight datasets, ASM with FedGEM attains a mean ARI of 0.333, against 0.256 for the next best label-free estimator and 0.361 when the true local counts are supplied. It also gives the most reliable global estimates of $K$ and is robust when client size is decoupled from local cardinality.