Self-Supervised Consistency Regularization for Poisoning-Resilient Memory Sharing in Multi-Agent Networks
American Journal Of Big Data · 01.04.2026 · 7 (1) · 229-239
DOI: 10.71465/ajbd3657 ↗Kurz gefasst
Fachartikel in der Zeitschrift American Journal Of Big Data, online erschienen am 01.04.2026. Verfasst von Jeroen van Dijk und 2 weiteren Autor:innen. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0).
Abstract (Originalsprache)
Collaborative agents frequently rely on shared memory states without external supervision. This study proposes a self-supervised consistency regularization mechanism to detect and correct poisoned memory updates. Each agent maintains a local predictive model to estimate expected peer memory updates. Deviations are penalized using a consistency loss computed over sliding windows of interactions. The framework was evaluated on a 75-agent distributed planning benchmark under poisoning rates ranging from 5% to 25%. At a 20% injection rate, baseline systems experienced a 36.7% performance drop, whereas consistency-regularized systems limited degradation to 14.2%. False-positive rejection rate remained below 7.1%. Self-supervised consistency constraints enhance collaborative robustness without centralized validation.
Zitieren
APA
Dijk, J. V., Vries, M. D., & Meer, S. V. D. (2026). Self-Supervised Consistency Regularization for Poisoning-Resilient Memory Sharing in Multi-Agent Networks. American Journal Of Big Data, 7(1), 229-239. https://doi.org/10.71465/ajbd3657
Vancouver
Dijk JV, Vries MD, Meer SVD. Self-Supervised Consistency Regularization for Poisoning-Resilient Memory Sharing in Multi-Agent Networks. American Journal Of Big Data. 2026;7(1):229-239. doi:10.71465/ajbd3657
BibTeX
@article{Dijk2026,
author = {Dijk, Jeroen van and Vries, Maarten de and Meer, Sophie van der},
title = {Self-Supervised Consistency Regularization for Poisoning-Resilient Memory Sharing in Multi-Agent Networks},
journal = {American Journal Of Big Data},
year = {2026},
volume = {7},
number = {1},
pages = {229--239},
doi = {10.71465/ajbd3657}
}
EUVIMED