HDP-book Exercise 2.2.8
高维概率 练习 2.2.8 讨论
题目
Exercise 2.2.8 (Boosting randomized algorithms). Imagine we have an algorithm for solving some decision problem (e.g. is a given number p p p a prime?). Suppose the algorithm makes a decision at random and returns the correct answer with probability 1 2 + δ \frac{1}{2} + \delta 21+δ with some δ > 0 \delta > 0 δ>0, which is just a bit better than a random guess. To improve the performance, we run the algorithm N N N times and take the majority vote. Show that, for any ϵ ∈ ( 0 , 1 ) \epsilon \in (0, 1) ϵ∈(0,1), the answer is correct with probability at least 1 − ϵ 1 - \epsilon 1−ϵ, as long as
N ≥ 1 2 δ 2 ln ( 1 ϵ ) N \ge \dfrac{1}{2\delta^2} \ln \left( \dfrac{1}{\epsilon} \right) N≥2δ21ln(ϵ1)
Hint: Apply Hoeffding’s inequality for X i X_i Xi being the indicators of the wrong answers.
简单翻译一下就是:随机猜,每一次猜对的概率为 1 2 + δ \frac{1}{2} + \delta 21+δ。 需要猜多少次才能确保猜对的概率至少是 1 − ϵ 1 - \epsilon 1−ϵ? 提示是考虑猜错的情况。
预备知识
Theorem 2.2.6 (Hoeffding’s inequality for general bounded random variables). Let X 1 , ⋯ , X N X_1, \cdots, X_N X1,⋯,XN be independent random variables. Assume that