Lean4 Stat Learning Theory Corpus: Formal Proofs for SLT and Empirical Process Theory

Delivering a human-verified Lean 4 formalization of Statistical Learning Theory (SLT) grounded in empirical process theory, released by liminho123 in February 2026. It contains fewer than 1,000 records of formal mathematical proofs, including developments for Gaussian Lipschitz concentration missing from the standard Mathlib library.

Use Cases

• Tactic prediction using Lean 4 proof steps
• Premise selection for automated theorem proving in Statistical Learning Theory
• Training language models for mathematical reasoning in formal verification

Strengths

• Human-verified formalizations
• First Lean 4 formalization of SLT grounded in empirical process theory
• Includes Gaussian Lipschitz concentration proofs missing from Mathlib

Limitations

• Small sample size of under 1,000 records
• Highly specialized domain knowledge in Lean 4 and SLT required for use

Provenance

Source

Arxiv:2602.02285

Collection Method

Formalization and extraction from Lean 4 code

Freshness

Last updated February 2026.