Prove2Me
Navigate
DiscoverCollectionsFormalpediaGalaxyBlogsUsersMomentumMy Missions+
Prove2Me
Log in
← Formalpedia

p. 356 — for J = 2 Algorithm 6 is Algorithm 3: centred F, p = e^F/(e^F + e^{−F}), F ← F + ½ E_w(z|x)

Proved
AddLogReg.Multiclass.two_class_reduces

by mikedeng1 · Oct 6, 2026 · Mathlib 0df444a (Lean v4.33.1)

logistic-regressionlogitboostmulticlass-classificationp2o-batch-pfp2ap2o-gran-per-chapterp2o-plan-paperp2o-v1

Take J=2J = 2J=2, a probability measure ν\nuν on X×{1,2}X\times\{1,2\}X×{1,2}, a fit F(x)=(F1(x),F2(x))F(x) = (F_1(x),F_2(x))F(x)=(F1​(x),F2​(x)) and a point xxx with F1(x)+F2(x)=0F_1(x)+F_2(x) = 0F1​(x)+F2​(x)=0. Then, in the population version of Algorithm 6:

  1. the model probability of class 1 is the two-class symmetric logistic (30):
p1(x)=eF1(x)eF1(x)+e−F1(x);p_1(x) = \frac{e^{F_1(x)}}{e^{F_1(x)}+e^{-F_1(x)}};p1​(x)=eF1​(x)+e−F1​(x)eF1​(x)​;
  1. the updated score of class 1 is
F1(x)+12 Ew1(z1∣x),z1=y1∗−p1(x)p1(x)(1−p1(x)),w1=p1(x)(1−p1(x)),F_1(x) + \tfrac12\,E_{w_1}(z_1\mid x),\qquad z_1 = \frac{y^*_1-p_1(x)}{p_1(x)(1-p_1(x))},\quad w_1 = p_1(x)(1-p_1(x)),F1​(x)+21​Ew1​​(z1​∣x),z1​=p1​(x)(1−p1​(x))y1∗​−p1​(x)​,w1​=p1​(x)(1−p1​(x)),

which is Algorithm 3's update F(x)←F(x)+12fm(x)F(x)\leftarrow F(x)+\frac12 f_m(x)F(x)←F(x)+21​fm​(x) with class 1 as y=1y=1y=1; 3. the update stays centred: the new F2(x)F_2(x)F2​(x) is minus the new F1(x)F_1(x)F1​(x).

This makes precise the sentence "Algorithm 6 is a natural generalization of Algorithm 3 for fitting the J-class logistic regression model (40)".

Formalization Note Classes are Fin 2; Algorithm 3's step is written out here, not imported.

Preamble
import Mathlib
import Definitions.Def_AddLogReg_Multiclass_Setting

open MeasureTheory ProbabilityTheory ENNReal
Formal statement
namespace AddLogReg.Multiclass

theorem two_class_reduces {X : Type*} [MeasurableSpace X]
    (ν : Measure (X × Fin 2)) [IsProbabilityMeasure ν] (F : X → Fin 2 → ℝ) (x : X)
    (hF : F x 0 + F x 1 = 0) :
    softmax (F x) 0 = Real.exp (F x 0) / (Real.exp (F x 0) + Real.exp (-F x 0)) ∧
      logitBoostJStep ν F x 0 =
        F x 0 + (1 / 2) * wCondExpBy ν (fun x' _ => classWeight F x' 0)
          (fun x' y => workingResponse F x' y 0) x ∧
      logitBoostJStep ν F x 1 = -logitBoostJStep ν F x 0 := by sorry

end AddLogReg.Multiclass
Source
Friedman, Hastie and Tibshirani, Additive logistic regression: a statistical view of boosting, Ann. Statist. 28 (2000), p. 356, sentence before Result 6; p. 351, Algorithm 3 and (30)
Human review
  • Endorsed by Shuze Chen · Oct 7, 2026

    Confirmed by the moderator at approval.

  • Endorsed by mikedeng1 · Oct 7, 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