VertCoHiRF: Decentralized Vertical Clustering Beyond k-means

2026-02-07

AI Summary

Existing vertical federated clustering methods rely on centralized coordination or the exchange of feature statistics, making them vulnerable to view heterogeneity, adversarial behavior, and insufficient privacy guarantees. This work proposes a fully decentralized vertical federated clustering framework in which participants perform local clustering independently and then reach consensus solely through ordinal rankings at the sample identifier level to construct a shared hierarchical clustering structure. By eliminating the need for feature statistic exchange or noise injection, the approach communicates only sample IDs, labels, and ranking information, thereby supporting heterogeneous feature partitions and arbitrary local clustering algorithms. The proposed decentralized ordinal consensus mechanism, integrated with the Cluster Fusion Hierarchy, effectively aligns cross-view clustering results while ensuring built-in privacy, low communication overhead, strong robustness, and competitive clustering performance.

Technology Category

• Machine Learning: Clustering
• Search and Optimization: Distributed Search
• Constraint Satisfaction and Optimization: Distributed CSP/Optimization

Application Category

• Systems and Infrastructure for Web, Mobile and WoT: Decentralized Web and Fediverse systems
• User Modeling, Personalization and Recommendation: Federated recommendation systems and personalization
• Security and Privacy: Data transparency and provenance

Abstract

Vertical Federated Learning (VFL) enables collaborative analysis across parties holding complementary feature views of the same samples, yet existing approaches are largely restricted to distributed variants of $k$-means, requiring centralized coordination or the exchange of feature-dependent numerical statistics, and exhibiting limited robustness under heterogeneous views or adversarial behavior. We introduce VertCoHiRF, a fully decentralized framework for vertical federated clustering based on structural consensus across heterogeneous views, allowing each agent to apply a base clustering method adapted to its local feature space in a peer-to-peer manner. Rather than exchanging feature-dependent statistics or relying on noise injection for privacy, agents cluster their local views independently and reconcile their proposals through identifier-level consensus. Consensus is achieved via decentralized ordinal ranking to select representative medoids, progressively inducing a shared hierarchical clustering across agents. Communication is limited to sample identifiers, cluster labels, and ordinal rankings, providing privacy by design while supporting overlapping feature partitions and heterogeneous local clustering methods, and yielding an interpretable shared Cluster Fusion Hierarchy (CFH) that captures cross-view agreement at multiple resolutions.We analyze communication complexity and robustness, and experiments demonstrate competitive clustering performance in vertical federated settings.

Problem

• Vertical Federated Learning
• Decentralized Clustering
• Heterogeneous Views
• Robustness
• Privacy

Innovation

• Vertical Federated Learning
• Decentralized Clustering
• Structural Consensus
• Cluster Fusion Hierarchy
• Privacy-by-Design