No label defined (Q4942227)

The authors review techniques in obtaining upper bounds on the tail probability of the type \(P(S_n- E(S_n)\geq na)\) where \(S_n\) is a sum of \(n\) independent random variables \(X_i\) leading to \textit{H. Chernoff's} inequality [Ann. Math. Stat. 23, 493-507 (1952; Zbl 0048.11804)], \textit{W. Hoeffding's} inequality [J. Am. Stat. Assoc. 58, 13-30 (1963; Zbl 0127.10602)] and \textit{G. Bennett's} inequality [ibid. 57, 33-45 (1962; Zbl 0104.11905)]. They provide a systematic account of a general method to obtain upper bounds of tail probabilities. New proofs of the Hoeffding and Bennett bounds are given and a generalization of these inequalities which take into account the first \(k\) moments of \(X_i\) are discussed. An application to problems in computer science is given. For the entire collection see [Zbl 0930.00017].