Cross validation refers to a set of statistical methods for the assessment of machine learning models or variables and features inside them. There are several variations on these methods, e.g. k-fold or leave-p-out cross validation. Generally if a method is called by a number e.g. four-fold cross validation, then it is an example of a k-fold method. Practically speaking, usually cross validation is used to overcome the potential bias in test and training data sets to get a well fit model; however it can be used in feature selection as well.