introvoyz045's picture
Migrated from GitHub
4a56cd6 verified
Raw History Blame Contribute Delete
1.74 kB
import OAI.Probability.ProjectionMoments.Model
namespace OAI
/-!
Finite-state Gaussian regression with noiseless labels and angular accuracy.
-/
open MeasureTheory ProbabilityTheory Set Filter
open scoped ENNReal NNReal RealInnerProductSpace
noncomputable section
universe u
namespace NoiselessRegression
namespace Learner
def success {d M T : β„•} {Ξ© : Type u} [MeasurableSpace Ξ©]
(L : Learner d M T Ξ©) (ρ : Measure Ξ©) (Ξ΅ : ℝ) (s : UnitSphere d) : ℝβ‰₯0∞ :=
(ρ.prod (sampleLaw d T))
{z | angularError s (L.estimate s z.1 z.2) ≀ Ξ΅}
end Learner
def SubquadraticMemory : Prop :=
βˆƒ c : ℝ, 0 < c ∧ βˆ€ M : β„• β†’ β„•,
Asymptotics.IsLittleO atTop (fun d => (M d : ℝ)) (fun d => (d : ℝ) ^ 2) β†’
βˆƒ dβ‚€ : β„•, βˆ€ d : β„•, dβ‚€ ≀ d β†’ βˆ€ T : β„•, βˆ€ Ξ΅ : ℝ,
0 < Ξ΅ β†’ Ξ΅ ≀ 1 / 10 β†’
βˆ€ (Ξ© : Type u) [MeasurableSpace Ξ©] (ρ : Measure Ξ©), IsProbabilityMeasure ρ β†’
βˆ€ L : Learner d (M d) T Ξ©, L.Admissible ρ β†’
(((2 : β„•) / (3 : β„•) : ℝβ‰₯0∞) ≀ L.uniformSuccess ρ Ξ΅ ∨
βˆ€ s, ((2 : β„•) / (3 : β„•) : ℝβ‰₯0∞) ≀ L.success ρ Ξ΅ s) β†’
c * (d : ℝ) * Real.log (1 / Ξ΅) ≀ (T : ℝ)
def FixedQuadraticMemory : Prop :=
βˆ€ Aβ‚€ : ℝ, 0 < Aβ‚€ β†’ βˆƒ c : ℝ, 0 < c ∧ βˆƒ dβ‚€ : β„•,
βˆ€ d : β„•, dβ‚€ ≀ d β†’ βˆ€ M T : β„•, (M : ℝ) ≀ Aβ‚€ * (d : ℝ) ^ 2 β†’
βˆ€ Ξ΅ : ℝ, 0 < Ξ΅ β†’ Ξ΅ ≀ 1 / 10 β†’
βˆ€ (Ξ© : Type u) [MeasurableSpace Ξ©] (ρ : Measure Ξ©), IsProbabilityMeasure ρ β†’
βˆ€ L : Learner d M T Ξ©, L.Admissible ρ β†’
((2 : β„•) / (3 : β„•) : ℝβ‰₯0∞) ≀ L.uniformSuccess ρ Ξ΅ β†’
c * (d : ℝ) * Real.log (1 / Ξ΅) ≀ (T : ℝ)
end NoiselessRegression
end
end OAI