Boosting Decision Trees

• How can we train a decision tree that has errors?
  • Pruning
  • Cross-validation
• How can we guarantee errors?
• Given an error-prone decision tree, the algorithm is the same as with decision stumps