stat.MLOct 28, 2026

CausalGrade: Evidence-Guided Causal Graph Scoring with Jev

Authors: Jiangming Zhong, Jiawei Zhu, Guanting Wu, Qiaoxi Cai, Wanyi Chen, Jiale Li

Organizations: Guangdong University of Technology, Guangzhou, China

Abstract

Evaluating the structural quality of candidate causal graphs when the true causal structure is unknown is an important problem in applying causal discovery results. This paper proposes CausalGrade, a causal graph scoring method based on the probabilistic judgment model Jev. The method jointly provides Jev with evidence of association, conditional dependence, and directionality from observational data and edge probabilities produced by the causal discovery foundation model CDFM. Jev produces probabilistic judgments about direct causal edges, and a graph-level score is obtained by measuring the agreement between the adjacency structure of the graph under evaluation and these edge-level judgments. Across 300 synthetic datasets, the score exhibits a relatively strong negative correlation with structural error in all 20 combinations of four causal discovery algorithms and five variable counts, indicating that higher scores generally correspond to smaller structural errors. In 500 synthetic graph selection tasks, CausalGrade achieves a best-candidate hit rate of 54.4%, exceeding the 48.4% achieved by the Self-Compatibility metric κGκ_G, further supporting its ability to discriminate the structural quality of graphs under evaluation.

Figures & tables

Explore similar work

May 19, 2026

Score-Based Causal Discovery of Latent Variable Causal Models

Identifying latent variables and the causal structure involving them is essential across various scientific fields. While many existing works fall under the category of constraint-based methods (with e.g. conditional independence or rank deficiency tests), they may face empirical challenges such as testing-order dependency, error propagation, and choosing an appropriate significance level. These issues can potentially be mitigated by properly designed score-based methods, such as Greedy Equivalence Search (GES) (Chickering, 2002) in the specific setting without latent variables. Yet, formulating score-based methods with latent variables is highly challenging. In this work, we develop score-based methods that are capable of identifying causal structures containing causally-related latent variables with identifiability guarantees. Specifically, we show that a properly formulated scoring function can achieve score equivalence and consistency for structure learning of latent variable causal models. We further provide a characterization of the degrees of freedom for the marginal over the observed variables under multiple structural assumptions considered in the literature, and accordingly develop both exact and continuous score-based methods. This offers a unified view of several existing constraint-based methods with different structural assumptions. Experimental results validate the effectiveness of the proposed methods.
Oct 15, 2026

Graph Dirichlet Energy for Fitting-Free Bivariate Causal Inference

We propose the Graph Dirichlet Energy Causal Inference (GDECI) criterion for distinguishing cause from effect in bivariate settings. For a candidate direction X→YX\to Y, we construct a kernel similarity graph from XX and treat YY as a graph signal. The score is the Dirichlet energy of this signal, induced by the unnormalized graph Laplacian. The reverse score swaps the variables, and the lower-energy direction is inferred as causal. The method requires neither regression fitting nor independence testing. After the empirical probability integral transform, its decisions are invariant under strictly monotone transformations of either variable, and the graph operations reduce to convolutions, allowing exact evaluation in O(nlog⁡n)O(n\log n) time and O(n)O(n) memory. At the population level, GDECI compares degree-weighted Bayes squared prediction errors plus a non-negative smoothness term of the regression function. As the bandwidth tends to zero, the weights become uniform and the smoothness term vanishes, reducing the criterion to a RECI-type comparison of expected conditional variances in copula space. Empirically, we compare GDECI with QCCD, GRCI, CAM, IGCI (two variants), RESIT, LOCI, QPE-k, and RECI (two variants), in accuracy on 1899 cause--effect pairs and in inference speed. GDECI ranked fifth of eleven methods in macro-averaged accuracy on 12 synthetic suites (0.7760.776). On the real-world Tuebingen pairs, it ranked third in unweighted accuracy (0.6840.684) and fifth in weighted accuracy (0.6780.678). It was also the fastest of the compared methods up to n≈104n\approx10^{4}. The accompanying code repository: https://github.com/povidanius/gdeci.
Jun 1, 2026

Consistency evaluation of benchmarks used for causal discovery

In graphical causal model, causal discovery aims to construct a causal graph based on numerical data and domain knowledge in plain text. However, the evaluation of causal discovery methods remains a challenge in the area as the progress of domain researches often makes benchmark causal graphs contain mis-aligned knowledge. This problem especially affects the evaluation of large language model (LLM) based causal discovery methods as they are sensitive to the new discoveries in the literature. This work is the first to systematically study the quality of benchmark causal graphs. Specifically, we design a pipeline that automatically retrieves relevant research papers from scientific databases, and prompts LLMs to check the consistency between the benchmark causal graphs and domain research papers. We evaluate 11 popular real-world benchmarks, for which our pipeline in total proceeds 38,081 domain papers. Our results show that popular benchmarks vary significantly in their consistency with domain research, with clear implications for causal discovery research.