Hoeffding Inequality

Let X1,X2,...,XkX_1,X_2,...,X_k be independent random variables with each Xi[ai,bi]X_i \in [a_i,b_i]. Let μi=𝔼[Xi]\mu_i=\mathbb{E}[X_i] and μ=iμi\mu=\sum_i \mu_i. Then, for any α>0\alpha > 0, S=iXiS = \sum_i X_i satisfies: Pr[|Sμ|>α]2eα2i=1k(biai)2\mathrm{Pr}[|S-\mu|>\alpha] \leq 2e^{-\frac{\alpha^2}{\sum_{i=1}^k (b_i-a_i)^2}}


Example of Concentration inequality.