A Prerequisite Random Variables

Many commonly used distributions satisfy this assumption, e.g., Bernoulli, uniform, and Gaussian We introduce a standard concentration bound for subgaussian random variables. Then for any > 0, it holds that P " Take Bernoulli distribution as an example. This will not change the argument. Let E be the event that the output is h . Then by Jensen's inequality, we have E To show Proposition 5, we need a standard concept--stopping time.