Power to the Clients: Federated Learning in a Dictatorship Setting

2025-10-25

AI Summary

This paper identifies a novel class of malicious clients in federated learning—“dictator clients”—capable of unilaterally nullifying all other clients’ model updates without accessing their private data, thereby monopolizing global model evolution. We formally define this attack model and systematically analyze its efficacy under complex multi-client interactions—including collusion, competition, and betrayal. Through rigorous convergence analysis and empirical evaluation on CV and NLP tasks, we demonstrate that dictator attacks can fully hijack the direction of global model updates, driving the model far from the optimal solution. The work exposes a critical security vulnerability in decentralized training and introduces the first theoretical framework for quantitatively characterizing the damage boundary of dictator attacks. This framework provides foundational support for designing robust federated learning mechanisms resilient to such centralized manipulation.

Technology Category

• Machine Learning: Distributed Machine Learning & Federated Learning
• Computer Vision: Adversarial Attacks & Robustness
• Multiagent Systems: Adversarial Agents

Application Category

• User Modeling, Personalization and Recommendation: Federated recommendation systems and personalization
• Security and Privacy: Large-scale security measurements
• Graph Algorithms and Modeling for the Web: Foundation models and LLMs for Web-related graphs

Abstract

Federated learning (FL) has emerged as a promising paradigm for decentralized model training, enabling multiple clients to collaboratively learn a shared model without exchanging their local data. However, the decentralized nature of FL also introduces vulnerabilities, as malicious clients can compromise or manipulate the training process. In this work, we introduce dictator clients, a novel, well-defined, and analytically tractable class of malicious participants capable of entirely erasing the contributions of all other clients from the server model, while preserving their own. We propose concrete attack strategies that empower such clients and systematically analyze their effects on the learning process. Furthermore, we explore complex scenarios involving multiple dictator clients, including cases where they collaborate, act independently, or form an alliance in order to ultimately betray one another. For each of these settings, we provide a theoretical analysis of their impact on the global model's convergence. Our theoretical algorithms and findings about the complex scenarios including multiple dictator clients are further supported by empirical evaluations on both computer vision and natural language processing benchmarks.

Problem

• Dictator clients can erase other clients' model contributions
• Attack strategies enable malicious control in federated learning
• Multiple dictator clients impact global model convergence theoretically

Innovation

• Introduces dictator clients to erase other contributions
• Proposes attack strategies for malicious client manipulation
• Analyzes multiple dictator collaboration and betrayal scenarios