Computer Science 8751
Advanced Machine Learning
Homework Assignment 3 (10 points)
Due October 28, 2003

  1. Mitchell, Page 388, Exercise 13.2
  2. One interesting question to examine regarding a dataset is what portions of the feature space are empty or nearly empty and therefore interesting. For example, it might be the case that three features are connected and are always high together or always low together. In this case, there would be no examples with high values of two of the features and low values for the other feature. Explain how the K-means clustering algorithm could be used to find portions of space with no examples. Indicate the types of situations your new algorithm would work for and where it would not work.
  3. Mitchell, Page 198, Exercise 6.1