Bootstrapping are techniques to start processes that otherwise have a "chicken and egg" nature. For example, in the {k-means algorithm}}, once one has a putatuve set of clusters new training items con be addedd to the nearest cluster gradually altering its centroid, but one needs a set of initial cluster centres for this -- one bootsrap approach to this is to chose k items randomly from the traing set as the initial cluster centroids. The term comes from the idea of lifting oneself up using one's own "boot straps' (or shoe laces).
Used on Chap. 10: page 209; Chap. 11: page 238; Chap. 14: page 331; Chap. 18: page 439
Also known as bootstrap, bootstrapped