Causal Inference–Driven Repair of Memory Poisoning in Multi-Agent Collaborative Systems
American Journal of Machine Learning · 01.04.2026 · 7 (1) · 155-162
DOI: 10.71465/ajml3660 ↗Kurz gefasst
Fachartikel in der Zeitschrift American Journal of Machine Learning, online erschienen am 01.04.2026. Verfasst von Liam Anderson und 2 weiteren Autor:innen. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0).
Abstract (Originalsprache)
Memory poisoning in multi-agent systems may distort downstream decision processes by altering causal dependencies among shared states. This study introduces a structural causal model (SCM) to identify and repair corrupted memory segments. Agent memory variables are modeled as nodes in a directed acyclic graph learned through constraint-based causal discovery. Intervention tests estimate causal effect shifts caused by adversarial perturbations. Corrupted nodes are reconstructed using do-calculus–based counterfactual inference. Experiments were conducted on a 90-agent distributed task-planning simulation with injected poisoning rates of 10%–30%. Causal repair reduced downstream decision error from 34.6% to 12.8% under a 20% poisoning scenario. Recovery latency decreased by 41.3% compared with naive memory reset strategies. Causal modeling provides interpretable and effective repair mechanisms for collaborative agent memory integrity
Zitieren
APA
Anderson, L., Taylor, E., & Wilson, J. (2026). Causal Inference–Driven Repair of Memory Poisoning in Multi-Agent Collaborative Systems. American Journal of Machine Learning, 7(1), 155-162. https://doi.org/10.71465/ajml3660
Vancouver
Anderson L, Taylor E, Wilson J. Causal Inference–Driven Repair of Memory Poisoning in Multi-Agent Collaborative Systems. American Journal of Machine Learning. 2026;7(1):155-162. doi:10.71465/ajml3660
BibTeX
@article{Anderson2026,
author = {Anderson, Liam and Taylor, Emily and Wilson, Jack},
title = {Causal Inference–Driven Repair of Memory Poisoning in Multi-Agent Collaborative Systems},
journal = {American Journal of Machine Learning},
year = {2026},
volume = {7},
number = {1},
pages = {155--162},
doi = {10.71465/ajml3660}
}
EUVIMED