Implement basic k-NN classification and the condensed 1-NN algorithm for the Letter Recognition Dataset. The first 15,000 examples are for training and the remaining 5,000 for testing. What is the purpose of using the condensed 1-NN algorithm in k-NN classification?
1) To reduce the number of training examples by selecting only the most representative ones
2) To increase the accuracy of the k-NN classification algorithm
3) To speed up the computation time of the k-NN classification algorithm
4) To improve the generalization ability of the k-NN classification algorithm