analiza aiqbench
FedNIA introduces a novel defense mechanism for federated learning by using noise-induced activation analysis to detect data poisoning without requiring a central test dataset. This approach effectively identifies adversarial clients across various attack vectors, including backdoors and label flipping, even in non-iid environments with multiple malicious nodes. The method offers a practical pathway to enhance the security and robustness of distributed AI systems against collaborative threats.
FedNIA: Noise-Induced Activation Analysis for Mitigating Data Poisoning in Federated Learning
arXiv:2502.16396v2 Announce Type: replace-cross Abstract: Federated learning systems are increasingly threatened by data poisoning attacks, where malicious clients compromise global models by contributing tampered updates. Existing defenses often rely on impractical assumptions, such as access to a central test dataset, or fail to generalize across diverse attack types, particularly those involving multiple malicious clients working collaboratively. To address this, we propose Federated Noise-Induced Activation Analysis (FedNIA), a novel defense framework to identify and exclude adversarial clients without relying on any central test dataset. FedNIA injects random noise inputs to analyze the layerwise activation patterns in client models leveraging an autoencoder that detects abnormal behaviors indicative of data poisoning. FedNIA can defend against diverse attack types, including sample poisoning, label flipping, and backdoors, even in scenarios with multiple attacking nodes. Experimental results on non-iid federated datasets demonstrate its effectiveness and robustness, underscoring its potential as a foundational approach for enhancing the security of federated learning systems.
Abstract:Federated learning systems are increasingly threatened by data poisoning attacks, where malicious clients compromise global models by contributing tampered updates. Existing defenses often rely on impractical assumptions, such as access to a central test dataset, or fail to generalize across diverse attack types, particularly those involving multiple malicious clients working collaboratively. To address this, we propose Federated Noise-Induced Activation Analysis (FedNIA), a novel defense framework to identify and exclude adversarial clients without relying on any central test dataset. FedNIA injects random noise inputs to analyze the layerwise activation patterns in client models leveraging an autoencoder that detects abnormal behaviors indicative of data poisoning. FedNIA can defend against diverse attack types, including sample poisoning, label flipping, and backdoors, even in scenarios with multiple attacking nodes. Experimental results on non-iid federated datasets demonstrate its effectiveness and robustness, underscoring its potential as a foundational approach for enhancing the security of federated learning systems.
Machine Learning (cs.LG); Artificial Intelligence (cs.AI); Cryptography and Security (cs.CR)
arXiv:2502.16396 [cs.LG]
(or arXiv:2502.16396v2 [cs.LG] for this version)
https://doi.org/10.48550/arXiv.2502.16396 Focus to learn more arXiv-issued DOI via DataCite
https://doi.org/10.1109/TKDE.2026.3736353 Focus to learn more DOI(s) linking to related resources
arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website.
Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.
Have an idea for a project that will add value for arXiv's community? Learn more about arXivLabs.
Text integral sub licența deschisă a publicației · 259 cuvinte
Subiecte conexe
Noise cancellation keeps quantum simulations running longer
Quantum Zeitgeist · 12 oct. 2026
Research Note: Filtering Subversion-Relevant Information From Pretraining Data Is Feasible
LessWrong · 5 oct. 2026
GLM-5.3 Sometimes Considers How Train-Time Actions Shape Future Behavior
LessWrong · 15 oct. 2026
Surface ‘dirt’ shields quantum sensors from noise
Quantum Zeitgeist · 13 oct. 2026