Which of the following statements about boosting is false?

a. Boosting works best when using logistic regression models as the individual classifiers.

b. Boosting constructs a sequence of classifiers for which the training set for classifier j + 1 depends on the prediction errors of classifiers j on its training set.

c. Boosting predictions are based on the votes of individual classifiers in which each vote is weighted proportional to the corresponding classifier’s training set accuracy.

d. The number of classifiers in a boosting committee is a parameter that must be determined via cross-validation to prevent overfitting.