Which of the following statements about ensemble methods is false?

a. In a boosting ensemble, the training data for an individual model depends on the predictions of the previously constructed individual models.

b. A bagging ensemble generates a composite prediction by averaging individual predictions from a collection of unstable learners trained on slightly different sets of training data.

c. A random forest is a type of bagging ensemble with an additional mechanism that attempts to make the individual trees more independent of each other.

d. In boosting, individual models can be trained independently of each other in parallel.