Federated Learning Benchmark Reveals Vulnerabilities in Aggregation Methods

Researchers have developed a benchmark to evaluate federated aggregation methods under various attack scenarios, including model poisoning and backdoor attacks. The study analyzed five aggregation methods across five datasets and four conditions, finding that Trimmed Mean performed best in clean conditions, while Krum excelled under sign-flipping and Gaussian attacks. The analysis also identified potential discrepancies in the FedPARETO scaffold and the metric implementation for BadNets, suggesting limitations in current evaluation practices. AI

IMPACT Highlights potential weaknesses in federated learning defenses, prompting further research into robust aggregation techniques.

RANK_REASON The cluster contains a single academic paper detailing a new benchmark and analysis of federated learning methods. [lever_c_demoted from research: ic=1 ai=1.0]

Read on arXiv cs.LG →

• BadNets: Identifying Vulnerabilities in the Machine Learning Model Supply Chain
• FedPARETO
• Krum
• The Street View House Numbers Dataset
• truncated mean

• paper
• safety

AI-generated summary · Google Gemini · from 1 sources. How we write summaries →