LGFeb 22, 2021

Clustering Algorithm to Detect Adversaries in Federated Learning

arXiv:2102.10799v112 citations
Originality Incremental advance
AI Analysis

This addresses security vulnerabilities in federated learning for IoT devices, offering a practical solution with incremental improvements to existing detection methods.

The paper tackles the problem of adversaries injecting false gradients in federated learning for IoT intrusion detection systems, proposing a clustering-based detection method that boosts global model accuracy to 99% even with 40% adversaries.

In recent times, federated machine learning has been very useful in building intelligent intrusion detection systems for IoT devices. As IoT devices are equipped with a security architecture vulnerable to various attacks, these security loopholes may bring a risk during federated training of decentralized IoT devices. Adversaries can take control over these IoT devices and inject false gradients to degrade the global model performance. In this paper, we have proposed an approach that detects the adversaries with the help of a clustering algorithm. After clustering, it further rewards the clients for detecting honest and malicious clients. Our proposed gradient filtration approach does not require any processing power from the client-side and does not use excessive bandwidth, making it very much feasible for IoT devices. Further, our approach has been very successful in boosting the global model accuracy, up to 99% even in the presence of 40% adversaries.

Foundations

The foundational work for this paper's niche, ranked by how specifically the neighbourhood builds on it — not by global fame.

Your Notes