Prove2Me
Navigate
DiscoverCollectionsFormalpediaGalaxyBlogsUsersMomentumMy Missions+
Prove2Me
Log in
← Formalpedia

Derivation of Result 6, step 1, p. 357 — multilogit coordinates G_j = F_j − F_b with base class b give the probabilities (40) and G_j = log(p_j/p_b)

Proved
AddLogReg.Multiclass.base_probs

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

multiclass-classificationmultinomial-logistic-regressionp2o-batch-pfp2ap2o-gran-per-chapterp2o-plan-paperp2o-v1softmax

Let F=(F1,…,FJ)F=(F_1,\dots,F_J)F=(F1​,…,FJ​) be a real vector, let pj=eFj/∑k=1JeFkp_j = e^{F_j}/\sum_{k=1}^J e^{F_k}pj​=eFj​/∑k=1J​eFk​ be its probabilities (40), and fix a base class bbb. Put Gj=Fj−FbG_j = F_j - F_bGj​=Fj​−Fb​, so that Gb=0G_b = 0Gb​=0. Then for every j≠bj\ne bj=b,

eGj1+∑k≠beGk=pj,11+∑k≠beGk=pb,Gj=log⁡pjpb.\frac{e^{G_j}}{1+\sum_{k\ne b} e^{G_k}} = p_j,\qquad \frac{1}{1+\sum_{k\ne b}e^{G_k}} = p_b,\qquad G_j = \log\frac{p_j}{p_b}.1+∑k=b​eGk​eGj​​=pj​,1+∑k=b​eGk​1​=pb​,Gj​=logpb​pj​​.

So the multilogit parametrization Gj(x)=log⁡[P(yj∗=1∣x)/P(yb∗=1∣x)]G_j(x) = \log[P(y^*_j = 1\mid x)/P(y^*_b=1\mid x)]Gj​(x)=log[P(yj∗​=1∣x)/P(yb∗​=1∣x)], Gb(x)=0G_b(x)=0Gb​(x)=0, used in step 1 of the Derivation of Result 6, describes the same probabilities as the symmetric parametrization (40); this is what allows the score and Hessian of step 1 to be written in terms of pj(x)p_j(x)pj​(x).

Formalization Note The page takes b=Jb = Jb=J and remarks that the choice of base class is arbitrary; here bbb is any class. The statement is about the values at a fixed xxx.

Preamble
import Mathlib
import Definitions.Def_AddLogReg_Multiclass_Setting

open MeasureTheory ProbabilityTheory ENNReal
Formal statement
namespace AddLogReg.Multiclass

theorem base_probs {J : ℕ} (F : Fin J → ℝ) (b j : Fin J) (hj : j ≠ b) :
    Real.exp (F j - F b) / (1 + ∑ k ∈ Finset.univ.erase b, Real.exp (F k - F b)) = softmax F j ∧
      1 / (1 + ∑ k ∈ Finset.univ.erase b, Real.exp (F k - F b)) = softmax F b ∧
      Real.log (softmax F j / softmax F b) = F j - F b := by sorry

end AddLogReg.Multiclass
Source
Friedman, Hastie and Tibshirani, Additive logistic regression: a statistical view of boosting, Ann. Statist. 28 (2000), p. 357, Derivation of Result 6, step 1, first display (multilogit parametrization G_j(x))
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