AI Helps Solve Statistical Problem

FlorianGallwitz · x · 2026-07-15

AI has helped solve a major problem in statistics: how to control the False Discovery Rate (FDR) in multiple hypothesis testing.

The article reviews the BH procedure, introduced by Benjamini and Hochberg in 1995. It proved capable of controlling FDR under independent data, leading to its widespread use in high-throughput scientific fields like genomics, astronomy, and economics.

However, real-world data is often correlated, such as linkage disequilibrium in genetic variation data. Subsequent research has continuously tried to generalize the BH procedure to correlated data scenarios. The article mentions that Benjamini and Yekutieli (2001) also studied its validity under positive dependency.

Related event: GPT-5.6 Reportedly Found a BH Counterexample(8 posts)→

Original post →

More from AGI Musings

AGI Musings channel →