next up previous
Next: AdaBoost [3 P] Up: MLA_Exercises_160106 Previous: Leave-one-out Error Estimation [4


Weak-Learner with Weighting [3 P]

Boosting requires the weak learners to handle a weighting of the training instances. Describe in pseudo-code how you could implement this weighting directly in the following algorithms, i.e. without resampling of the training set:
a)
k-Nearest-Neighbour
b)
Linear Regression
c)
C4.5 (Hint: Define a suitable variation of information gain. You do not need to describe pruning.)



Pfeiffer Michael 2006-01-18