Residual-Based Causal Discovery via Pretrained In-Context Quantile Regression proposes Loscar, an extension of the RESIT framework to location-scale noise models (LSNMs). Loscar estimates conditional mean and scale using quantile regression, allowing pretrained tabular foundation models to serve as plug‑in estimators. Experiments on bivariate and multivariate causal‑discovery benchmarks show that in‑context quantile regression outperforms a feasible generalized least squares (FGLS) variant, especially when causes are multivariate.
arXiv:2609.16931v1 Announce Type: cross Abstract: We propose Low-Rank Quantile Surfaces (LRQS), a bivariate causal model in which, in the causal direction, an unknown monotone transformation of the c...
arXiv:2606. 18011v1 Announce Type: cross Abstract: Constraint-based causal discovery relies on repeated conditional independence tests, but fast nonparametric tests often sacrifice calibration, especially when variables depend on the conditioning set through nonlinear relationships.
arXiv:2511. 14441v2 Announce Type: replace-cross Abstract: To distinguish Markov equivalent graphs in causal discovery, it is necessary to restrict the structural causal model.
The paper introduces Structural Causal Bottleneck Models (SCBMs), a new class of structural causal models where high‑dimensional causal effects are mediated by low‑dimensional bottleneck variables. SCBMs allow mechanism‑specific, target‑dependent dimensionality reduction while remaining estimable with standard learning algorithms. The authors prove that bottleneck variables are identifiable up to bijection from observational data, and demonstrate experimentally that using bottlenecks improves identification and finite‑sample efficiency in causal effect estimation for instrumental variables, mediation, and backdoor adjustment.
The article discusses recent advances in learning directed acyclic graphs (DAGs) from observational data, highlighting the challenges of enforcing acyclicity and dealing with latent confounding or heterogeneous noise. It reviews the evolution from combinatorial search methods to continuous, score‑based approaches that use smooth acyclicity constraints, and introduces concomitant DAG estimation techniques that jointly infer sparse causal structure and exogenous noise levels to improve robustness under heteroscedasticity and distribution shifts. The tutorial also outlines future research directions such as online, nonlinear, and neural causal discovery.
arXiv:2607. 11508v1 Announce Type: cross Abstract: Causal discovery, the process of recovering underlying causal structures from observational data, is a fundamental pursuit across scientific disciplines.
arXiv:2606. 08196v1 Announce Type: cross Abstract: We study causal discovery from observational data when some variables are hidden and the data-generating process follows a location-scale noise model (LSNM).
The paper reviews modern causal discovery techniques that go beyond conditional independence, highlighting how distributional properties, such as asymmetries, conditional moments, and environmental variation, can identify causal structure. It distinguishes between identification sources, statistical estimators, and computational implementations, and discusses challenges in the presence of latent confounding, feedback, and cyclic models. The authors also outline consistency, uncertainty quantification, and algorithmic guarantees, and provide guidance for comparing methods and a research agenda for future work.
arXiv:2607. 11510v1 Announce Type: new Abstract: Causal discovery from observational tabular data remains fundamentally challenging, primarily due to the heterogeneity of underlying causal mechanisms and the high-dimensional combinatorial search space of Directed Acyclic Graphs (DAGs).
arXiv:2606. 17516v1 Announce Type: cross Abstract: Causal discovery from observational data remains challenging due to the need to recover directed structure and latent confounding without interventions.
arXiv:2609.36881v1 Announce Type: new Abstract: Causal foundation models (CFMs) pre-trained on data generated from various structural causal models (SCMs) have been proposed for estimating causal eff...
AdaPS-LiNGAM is a new method for causal discovery in linear non‑Gaussian acyclic models (LiNGAM) that addresses the problem of small sample sizes relative to the number of variables. The approach adapts the predecessor selection process by reconstructing each residual directly from the original observations using a sparsely chosen subset of previously ordered variables, rather than relying on repeated residualization which becomes degenerate when variables outnumber samples. Experiments on synthetic data show that AdaPS‑LiNGAM achieves accurate causal‑structure recovery in sample‑limited settings and its performance degrades more gradually as the sample size decreases.