Answers are grounded in this paper and cite their sources.
Consider a sequence of independent identically distributed (i.i.d.) random variablesXₗ,X₂, ⋯, Xₙand a distortion measured(Xᵢ,X̂ᵢ)on the estimatesX̂ᵢofXᵢ. Two descriptionsi(X)∈ \{1,2, ⋯ ,2^{nR₁\}andj(X)∈ \{1,2, ⋯,2^{nR₂\}are given of the sequenceX=(X₁, X₂, ⋯ ,Xₙ). From these two descriptions, three estimates(i(X)), X2(j(X)), andX̂O(i(X),j(X))are formed, with resulting expected distortionsE {1/n} ∑ⁿₖ₌₁ d(Xₖ, X̂ₘₖ)=Dₘ, m=0,1,2.We find that the distortion constraintsD₀, D₁, D₂are achievable if there exists a probability mass distributionp(x)p(x̂₁,x̂₂,x̂₀|x)withEd(X,x̂ₘ)≤ Dₘsuch thatR₁>I(X;X̂₁),R₂>I(X;X̂₂),whereI(·)denotes Shannon mutual information. These rates are shown to be optimal for deterministic distortion measures.
No takes yet. Share an insight, caveat, or question.
Gamal et al. (1982) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: