RSP-2026-0006 · Computer ScienceSubmittedv1 · round 1
A Reproducible Benchmark for Small-Sample Causal Discovery
Rita Ora-Nguyen · submitted Jul 5, 2026
- causal inference
- benchmark
- reproducibility
15% · Submitted
Open peer-reviewed research. Publication here is a review record, not an endorsement of clinical use — this material is not medical advice.
Abstract
Causal-discovery methods are often evaluated on idiosyncratic simulations. We propose a shared, versioned benchmark with realistic small-sample regimes, standardized metrics, and a leaderboard protocol designed to resist overfitting. We seed it with faithful reimplementations of nine established algorithms.
Manuscript — v1
Viewer not loading? Open the PDF directly
The Review Record
0 reviews on recordReviews are confidential until the editorial decision, then the full thread — every round — goes public and stays public.
No reviews on the public record yet. The band is warming up.