New TOFD Framework Enhances Split Federated Learning Against Poisoning Attacks

Researchers have developed a new framework called Target-Oriented Feature Decoupling (TOFD) to combat poisoning attacks in Split Federated Learning (SFL). TOFD operates in three stages: identifying potential attack targets through class-specific margin perturbation, purifying poisoned data samples, and using an adversarial guidance model to decouple and suppress attack influences during optimization. Experiments show TOFD consistently outperforms existing defenses across various attack scenarios, offering robust protection with low computational overhead. AI

IMPACT Enhances the security and reliability of collaborative AI training in privacy-sensitive applications.

RANK_REASON The cluster contains an academic paper detailing a new method for improving AI security. [lever_c_demoted from research: ic=1 ai=1.0]

Read on arXiv cs.AI →

• Margin Perturbation
• Split Federated Learning
• Target-Oriented Feature Decoupling

• paper
• safety

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