arXiv 5 Aug 2026 · Machine Learning
arXiv:2608.06427 · PDF · Extracted main text
Generative models can reproduce an observational distribution while encoding an incorrect causal structure. We study a sequential game in which a structural causal generator proposes observational and interventional distributions, while an adversarial experimentalist selects interventions intended to maximally falsify the generator. The discriminator is therefore not merely a real-versus-synthetic classifier: it is indexed by an intervention and tests whether the generator reproduces the corresponding post-intervention law. We introduce Adversarial Causal Intervention Falsification (ACIF), formulate oracle and implementable versions of the game, and distinguish three objects that are often conflated: observational fit, interventional equivalence over an admissible query class, and point identification of a structural causal model. For finite model and intervention classes, we prove: (i) an exact reduction of the adversarial objective to a worst-intervention integral probability metric; (ii) identification up to interventional equivalence, with point identification under a separating intervention family; (iii) existence of mixed-strategy equilibria; (iv) finite-sample uniform convergence and margin-based model-selection guarantees; and (v) a logarithmic elimination guarantee for a disagreement-driven sequential design under a balanced-separation condition. We also give a complete linear-Gaussian example in which two observationally indistinguishable causal directions are separated by a single well-chosen intervention. The framework clarifies what an adversarial causal discriminator can and cannot certify, and provides a principled bridge between causal generative modeling, active causal discovery, and experimental design.
appendix boundary found by appendix_command · 86% of the source is main text. Read the extracted text to check this.
The works this paper leans on most, across its whole bibliography — not restricted to papers in our corpus. Ranked by composite intensity, which combines how often a work is mentioned, how many sections mention it, and how much of that falls in the main text rather than the appendix.
| Reference | Intensity | Mentions | Sections | Main text | |
|---|---|---|---|---|---|
| 1 | Nowak, R. D (2011) The geometry of generalized binary search | 0.843 | 3 | 3 | 100% |
| 2 | Scherrer, N., Bilaniuk, O., Annadani, Y., Goyal, A., Schwab, P., Sch… (2021) Learning neural causal models with active interventions | 0.843 | 3 | 3 | 100% |
| 3 | Agrawal, R., Squires, C., Yang, K. D., Shanmugam, K., and Uhler, C (2019) ABCD-strategy: Budgeted experimental design for targeted causal structure discovery | 0.737 | 3 | 2 | 100% |
| 4 | Hauser, A. and Bühlmann, P (2012) Two optimal strategies for active learning of causal models from interventional data | 0.737 | 3 | 2 | 100% |
| 5 | Sriperumbudur, B. K., Fukumizu, K., Gretton, A., Schölkopf, B., and… (2012) On the empirical estimation of integral probability metrics | 0.737 | 3 | 2 | 100% |
| 6 | Tigas, P., Jesson, A., Gal, Y., Foster, A., and Bauer, S (2023) Differentiable multi-target causal Bayesian experimental design | 0.737 | 3 | 2 | 100% |
| 7 | Arjovsky, M., Chintala, S., and Bottou, L (2017) Wasserstein generative adversarial networks | 0.644 | 2 | 2 | 100% |
| 8 | Brouillard, P., Lachapelle, S., Lacoste, A., Lacoste-Julien, S., and… (2020) Differentiable causal discovery from interventional data | 0.644 | 2 | 2 | 100% |
| 9 | Drouin, A., Andrews, B., et al (2025) Adversarial causal tuning for realistic time-series generation | 0.644 | 2 | 2 | 100% |
| 10 | Goodfellow, I., Pouget-Abadie, J., Mirza, M., Xu, B., Warde-Farley,… (2014) Generative adversarial nets | 0.644 | 2 | 2 | 100% |
Showing the top 10 of 17 scored citations.