Cslib.MachineLearning.PACLearning.IsPACLearnerFor.antitone_C
theoremverifiedformalizationimported Lean
The deterministic PAC learner predicate is antitone in the concept class: a learner for a larger class C' is also a learner for any subclass C ⊆ C', since the agnostic benchmark optimalError _ C ≥ optimalError _ C' makes the error requirement easier.
Statement
set_option autoImplicit true in set_option relaxedAutoImplicit true in open MeasureTheory Set in open scoped ENNReal NNReal in variable {α : Type*} {β : Type*} [MeasurableSpace α] [MeasurableSpace β] in theorem Cslib.MachineLearning.PACLearning.IsPACLearnerFor.antitone_C {m : ℕ} {ε δ : Set.Ioo (0 : ℝ≥0) 1} {C C' : ConceptClass α β} (hC : C ⊆ C') {𝒟 : Set (Measure (α × β))} (h : IsPACLearnerFor m ε δ C' 𝒟) : IsPACLearnerFor m ε δ C 𝒟
- Formalizes
- CSLib, `Cslib.MachineLearning.PACLearning.IsPACLearnerFor.antitone_C`
- Lean proof
- By Samuel Schlesinger, from github.com/leanprover/cslib at
3951377e5a3f(Copyright (c) 2026 Samuel Schlesinger. All rights reserved, Apache-2.0). Compiled here against leanprover/lean4:v4.33.1, Mathlib v4.33.1; written for leanprover/lean4:v4.33.0.Cslib.MachineLearning.PACLearning.IsPACLearnerFor.antitone_C← Cslib.MachineLearning.PACLearning.IsPACLearnerFor.antitone_C (Cslib/MachineLearning/PACLearning/Defs.lean:244): verbatim
- Verified
- 08 Oct 2026
- Axioms
Classical.choiceQuot.soundpropext- Built on 6
- Cslib.MachineLearning.PACLearning.IsPACLearnerFor
IsPACLearnerFor m ε δ C 𝒟asserts that there exists a learner usingmsamples that is(ε, δ)-correct for the concept classCover… - Cslib.MachineLearning.PACLearning.errorThe *prediction error* (0-1 loss) of a hypothesis
hunder a joint distributionDonα × β, defined as the probability that the… - Cslib.MachineLearning.PACLearning.LabeledSampleA *labeled sample* of size
mover domainαwith label typeβis a finite sequence of(point, label)pairs. - Cslib.MachineLearning.PACLearning.optimalErrorThe *optimal error* of a concept class
Cunder a joint distributionD, defined as the infimum oferror D cover all concepts… - Cslib.MachineLearning.PACLearning.ConceptClassA *concept class* over domain
αwith label typeβis a set of functionsα → β. For binary classification (β = Bool), this is… - Cslib.MachineLearning.PACLearning.LearnerA *learner* using
msamples is a function that takes a labeled sample and produces a hypothesis (a function from the domain to the label…
- Cslib.MachineLearning.PACLearning.IsPACLearnerFor
- Used by 3
- Cslib.MachineLearning.PACLearning.IsPACLearnable.antitone_CPAC learnability is antitone in the concept class: a subclass of a learnable class is learnable.
- Cslib.MachineLearning.PACLearning.sampleComplexity_mono_CDeterministic sample complexity is monotone in the concept class under
⊆: a smaller class (weaker agnostic benchmark) requires no more… - Cslib.MachineLearning.PACLearning.IsPACLearnable.sampleComplexity_mono_C
sampleComplexity_mono_Cfor a learnable class (learnability at the *larger* classC'is the hypothesis).
Lean source view module on GitHub
Choose a tactic — a marked line — to see the goals Lean had before and after it, as in Lean’s infoview. Step with ← → or j k.
set_option autoImplicit true inset_option relaxedAutoImplicit true inopen MeasureTheory Set inopen scoped ENNReal NNReal invariable {α : Type*} {β : Type*} [MeasurableSpace α] [MeasurableSpace β] in/-- The deterministic PAC learner predicate is antitone in the concept class: a learner for a larger class `C'` is also a learner for any subclass `C ⊆ C'`, since the agnostic benchmark `optimalError _ C ≥ optimalError _ C'` makes the error requirement easier. -/theorem Cslib.MachineLearning.PACLearning.IsPACLearnerFor.antitone_C {m : ℕ} {ε δ : Set.Ioo (0 : ℝ≥0) 1} {C C' : ConceptClass α β} (hC : C ⊆ C') {𝒟 : Set (Measure (α × β))} (h : IsPACLearnerFor m ε δ C' 𝒟) : IsPACLearnerFor m ε δ C 𝒟 := by obtain ⟨A, hA⟩ := h refine ⟨A, fun D inst hD => le_trans (measure_mono ?_) (@hA D inst hD)⟩ intro S hS have h_opt : optimalError D C' ≤ optimalError D C := iInf_le_iInf_of_subset hC calc optimalError D C' + (↑ε.val : ℝ≥0∞) ≤ optimalError D C + ↑ε.val := by gcongr _ < error D (A S) := hS