Random sampling

A large portion of machine learning and deep learning contains random sampling in various forms. In this case, a training set of 3,000,000,000 elements will prove difficult to implement without random sampling.

Random sampling applies to many methods as described in this book: Monte Carlo, stochastic gradient descent, and many others. No matter what name the sampling takes, they share common concepts to various degrees depending on the size of the dataset. The following description fits the Monte Carlo philosophy.