Simulated data is sometimes used in machhine learning when there is eietherb insuffciuetn real dat or it is hard the gather wr soem reaosn. An example of the latter is trainng AI for autonomous car, it is useful to have exampels of people stepping off the curb in frint of the car, but not a good idea to ather the information forreal, sothat VR and games engines are used to create realistic sceanrios with lots of fariations. This example is were the simulated data is created from a model. In other cases real daat is used to prime the process, for example, generating lots of small perturbations from each real data item.
Used on page 495