New algorithm CREATE-IF-LATE counters "Pac-Man attack" in decentralized learning

Researchers have developed a new algorithm called CREATE-IF-LATE (CIL) to address a novel threat in decentralized learning known as the "Pac-Man attack." This attack involves a malicious node probabilistically terminating random walks (RWs) that visit it, leading to the gradual extinction of active RWs and halting the learning process without detection. CIL is designed to be a fully decentralized and resilient mechanism that ensures the self-creation of RWs, preventing their extinction even under Pac-Man attacks. Theoretical analysis indicates that CIL guarantees the non-extinction and boundedness of the RW population, and enables stochastic gradient descent to converge with a quantifiable deviation from the true optimum, experiencing at most a linear time delay. AI

IMPACT This research introduces a novel defense mechanism against specific adversarial attacks in decentralized learning systems, potentially improving their robustness.

RANK_REASON The cluster contains a research paper detailing a novel algorithm and theoretical analysis for decentralized learning. [lever_c_demoted from research: ic=1 ai=1.0]

Read on arXiv cs.LG →

• arXiv
• CREATE-IF-LATE
• Pac-Man attack
• stochastic gradient descent
• Xingran Chen

• paper
• other

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