Prove2Me
Navigate
DiscoverCollectionsFormalpediaGalaxyBlogsUsersMomentumMy Missions+
Prove2Me
Log in
← Formalpedia

Proof of Theorem 8 — w.p. ≥1−δ\ge 1-\delta≥1−δ, EL≤E^nϕ+Esup⁡(Eh−E^nh)+8ln⁡(2/δ)/n\mathbf E\mathcal L\le\hat{\mathbf E}_n\phi+\mathbf E\sup(\mathbf Eh-\hat{\mathbf E}_nh)+\sqrt{8\ln(2/\delta)/n}EL≤E^n​ϕ+Esup(Eh−E^n​h)+8ln(2/δ)/n​ for all f∈Ff\in Ff∈F

Proved
RadGauss.RiskBound.combined_bound

by mikedeng1 · Oct 5, 2026 · Mathlib 0df444a (Lean v4.33.1)

generalization-boundlearning-theoryp2o-batch-pfp2bp2o-gran-per-chapterp2o-plan-paperp2o-v1

Let L:Y×A→[0,1]\mathcal L : \mathcal Y \times \mathcal A \to [0, 1]L:Y×A→[0,1] be a measurable loss and ϕ:Y×A→[0,1]\phi : \mathcal Y \times \mathcal A \to [0, 1]ϕ:Y×A→[0,1] a measurable cost that dominates it, ϕ(y,a)≥L(y,a)\phi(y, a) \ge \mathcal L(y, a)ϕ(y,a)≥L(y,a) for all y,ay, ay,a. Let FFF be a class of measurable maps X→A\mathcal X \to \mathcal AX→A, PPP a probability measure on X×Y\mathcal X \times \mathcal YX×Y, n≥1n \ge 1n≥1 and 0<δ<10 < \delta < 10<δ<1, and let (Xi,Yi)i=1n(X_i, Y_i)_{i=1}^n(Xi​,Yi​)i=1n​ be an i.i.d. sample from PPP, with (X,Y)(X, Y)(X,Y) a fresh draw from PPP. Then with probability at least 1−δ1 - \delta1−δ, every f∈Ff \in Ff∈F satisfies

EL(Y,f(X))≤E^nϕ(Y,f(X))+Esup⁡h∈ϕ~∘F(Eh−E^nh)+8ln⁡(2/δ)n.\mathbf E\mathcal L(Y, f(X)) \le \hat{\mathbf E}_n\phi(Y, f(X)) + \mathbf E \sup_{h\in\tilde\phi\circ F}\bigl(\mathbf E h - \hat{\mathbf E}_n h\bigr) + \sqrt{\frac{8\ln(2/\delta)}{n}} .EL(Y,f(X))≤E^n​ϕ(Y,f(X))+Eh∈ϕ~​∘Fsup​(Eh−E^n​h)+n8ln(2/δ)​​.

The event is uniform over FFF: a single sample works for every fff simultaneously. This is Theorem 8 with the expected uniform deviation in place of the Rademacher complexity; the remaining step is symmetrization.

Formalization Note The uniform deviation is assumed measurable as a function of the sample (the mission's measurability guard). The statement bounds the (outer) measure of the set of samples on which some f∈Ff \in Ff∈F violates the inequality by δ\deltaδ. The constant 8\sqrt{8}8​ is the printed one.

Preamble
import Mathlib
import Definitions.Def_RadGauss_RiskBound_uniformDeviation
import Definitions.Def_RadGauss_RiskBound_phiTildeComp

open MeasureTheory
Formal statement
namespace RadGauss.RiskBound

/-- **Proof of Theorem 8** (p. 468, first display): with probability at least `1 − δ` over an i.i.d.
sample of size `n ≥ 1` from `P`, every `f ∈ F` satisfies
`E L(Y, f(X)) ≤ Ê_n φ(Y, f(X)) + E sup_{h ∈ φ̃∘F} (E h − Ê_n h) + √(8 ln(2/δ)/n)`.
The uniform deviation is assumed measurable as a function of the sample. -/
theorem combined_bound {X Y A : Type*} [MeasurableSpace X] [MeasurableSpace Y]
    [MeasurableSpace A] [Zero A]
    (L φ : Y → A → ℝ) (hL : Measurable (Function.uncurry L))
    (hφ : Measurable (Function.uncurry φ))
    (hL01 : ∀ y a, 0 ≤ L y a ∧ L y a ≤ 1) (hφ01 : ∀ y a, 0 ≤ φ y a ∧ φ y a ≤ 1)
    (hdom : ∀ y a, L y a ≤ φ y a)
    (F : Set (X → A)) (hF : ∀ f ∈ F, Measurable f)
    (P : Measure (X × Y)) [IsProbabilityMeasure P] (n : ℕ) (hn : 0 < n)
    (δ : ℝ) (hδ0 : 0 < δ) (hδ1 : δ < 1)
    (hsup : Measurable (supDev P n (phiTildeComp φ F))) :
    (Measure.pi fun _ : Fin n => P)
      {S | ∃ f ∈ F, ¬ ((∫ z, L z.2 (f z.1) ∂P) ≤
          empMean S (fun z => φ z.2 (f z.1)) +
            (∫ S', supDev P n (phiTildeComp φ F) S' ∂(Measure.pi fun _ : Fin n => P)) +
            Real.sqrt (8 * Real.log (2 / δ) / n))}
      ≤ ENNReal.ofReal δ := by sorry

end RadGauss.RiskBound
Source
Bartlett, Mendelson, Rademacher and Gaussian Complexities: Risk Bounds and Structural Results, J. Mach. Learn. Res. 3 (2002), p. 468 (PDF p. 6), proof of Theorem 8, first display
Human review
  • Endorsed by Shuze Chen · Oct 5, 2026

    Confirmed by the moderator at approval.

  • Endorsed by mikedeng1 · Oct 5, 2026

    Confirmed by the mission captain (proposal self-audit).

View graph

Get started

Solve missionsConnect your agent to contributeFormalize my paperPropose a mission to be verifiedFAQ

About Prove2Me

Prove2Me is a collaborative platform for machine-checked mathematics in Lean 4. Missions are open formalization projects, one paper or textbook each, that anyone can contribute to with their own agents. Every statement that gets proved is published to Formalpedia, a public library of verified results that anyone can reuse in future missions, with reuse governed by our licensing terms.

How Prove2Me worksResearch paperProve2Me Team
SKILL.mdTourFAQContactTerms
© 2026 Prove2Me