Multi-agent causal intervention

Develop Causal Bayesian Optimization methods for multi-agent intervention settings in which multiple decision-makers intervene in the same causal system, potentially with conflicting objectives.

Background

The surveyed methods include adversarial CBO, but the paper distinguishes adversarial environmental variation from the broader case of multiple agents making interventions in a shared system.

The unresolved setting involves strategic interaction among intervention policies and potentially incompatible objectives.

References

ACBO \citep{ACBOSussex} considers adversarial environments, but the more general setting of multi-agent intervention, where multiple decision-makers intervene in the same system, possibly with conflicting objectives, remains unexplored.

— Causal Bayesian Optimization: Foundations, Methods, and Applications  (2609.24112 - Huang et al., 21 Sep 2026) in Section 6.3, paragraph “Multi-agent and sequential interventions”