No-Free-Lunch table𝒳 = {x₁, x₂, x₃}, uniform distribution, labels ±1, all 2³ = 8 functions
true function f*: training points: