Research map / Federated learning and privacy / Privacy attacks and secure learning

Robust and Scalable Collaborative Learning via Pull-Based Epidemic Communication

NeurIPS 2026 · Federated learning and privacy › Privacy attacks and secure learning · approach: Membership inference and auditing

Read the paperCheck your idea against 108,967 papers

AI-extracted summary. Written by AI from the paper's abstract only. Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Proposes Robust Pull-based Epidemic Learning (RPEL), a fully decentralized collaborative learning framework that achieves Byzantine robustness with lower communication cost than O(n^2) methods.

Problem

Collaborative machine learning is vulnerable to adversarial (Byzantine) behaviors during training, and existing defenses rely on central coordination or incur high communication costs that grow as O(n^2) with the number of nodes.

Approach

Uses pull-based epidemic communication, in which each node pulls models from a sampled subset of peers rather than all-to-all exchange, combined with robust aggregation, to achieve robustness without a central server and with communication scaling better than traditional methods.

Experimental setup

Builds on: epidemic learning

Closest papers by meaning

• Unified Breakdown Analysis for Byzantine Robust Gossip (ICML 2025 · similarity 0.87)
• On the Robustness of Distributed Machine Learning Against Transfer Attacks (AAAI 2025 · similarity 0.86)
• RESIST: Resilient Decentralized Learning Using Consensus Gradient Descent (NeurIPS 2026 · similarity 0.85)
• On the Tension between Byzantine Robustness and No-Attack Accuracy in Distributed Learning (ICML 2025 · similarity 0.84)
• Byzantine-Robust Decentralized Learning via Remove-then-Clip Aggregation (AAAI 2024 · similarity 0.84)
• Enhancing Robustness of Graph Neural Networks on Social Media with Explainable Inverse Reinforcement Learning (NeurIPS 2024 · similarity 0.84)