Biased halfspaces, noise sensitivity, and relative Chernoff inequalities (extended version)

In analysis of Boolean functions, a halfspace is a function f : {−1, 1}n → {0, 1} of the form f(x) = 1{a · x > t}, where ∑ i a i = 1. We show that if f is a halfspace with E[f ] = ǫ, then the degree-1 Fourier weight of f is W (f) = Θ(ǫ log(1/ǫ)), and the maximal influence of f is Imax(f) = Θ(ǫmin(1, a ′ √ log(1/ǫ))), where a = maxi |ai|. These results, which determine the exact asymptotic order of W (f) and Imax(f), provide sharp generalizations of theorems proved by Matulef, O’Donnell, Rubinfeld, and Servedio, and settle a conjecture posed by Kalai, Keller and Mossel. In addition, we present a refinement of the definition of noise sensitivity which takes into consideration the bias of the function, and show that (like in the unbiased case) halfspaces are noise resistant, and on the other hand, any noise resistant function is well-correlated to a halfspace. Our main tools are ‘relative’ forms of the classical Chernoff inequality, like the following one: Let {xi} be independent random variables uniformly distributed in {−1, 1}, and let ai ∈ R≥0 be such that ∑ i a i = 1. If for some t ≥ 0 we have Pr [∑ i aixi > t] = ǫ, then Pr[ ∑ i aixi > t+ δ] ≤ ǫ 2 holds for δ ≤ c/ √ log(1/ǫ), where c is a universal constant.