Talagrand's concentration inequality

From testwiki
Jump to navigation Jump to search

Template:Short description Template:Technical In the probability theory field of mathematics, Talagrand's concentration inequality is an isoperimetric-type inequality for product probability spaces.[1][2] It was first proved by the French mathematician Michel Talagrand.[3] The inequality is one of the manifestations of the concentration of measure phenomenon.[2]

Roughly, the product of the probability to be in some subset of a product space (e.g. to be in one of some collection of states described by a vector) multiplied by the probability to be outside of a neighbourhood of that subspace at least a distance t away, is bounded from above by the exponential factor e−t2/4. It becomes rapidly more unlikely to be outside of a larger neighbourhood of a region in a product space, implying a highly concentrated probability density for states described by independent variables, generically. The inequality can be used to streamline optimisation protocols by sampling a limited subset of the full distribution and being able to bound the probability to find a value far from the average of the samples.[4]

Statement

The inequality states that if Ω=Ω1×Ω2×⋯×Ωn is a product space endowed with a product probability measure and A is a subset in this space, then for any t≥0

Pr⁡[A]⋅Pr⁡[Atc]≤e−t2/4,

where Atc is the complement of At where this is defined by

At={x∈Ω:ρ(A,x)≤t}

and where ρ is Talagrand's convex distance defined as

ρ(A,x)=maxα,‖α‖2≤1 miny∈A ∑i:xi≠yiαi

where α∈𝐑n, x,y∈Ω are n-dimensional vectors with entries αi,xi,yi respectively and ‖⋅‖2 is the ℓ2-norm. That is,

‖α‖2=(∑iαi2)1/2

References

Template:Reflist


Template:Probability-stub