Computer Science 8751
Advanced Machine Learning
Homework Assignment 3 (20 points)
Due March 23, 2009

  1. A machine learning problem has four input features: A, which has possible values A1, A2 and A3; B, which has possible values of yes and no, C which can take on any value from 0.0 to 150.0, and D which has possible values of true, false or unknown. The output class to be calculated has four possible values: tiny, small, medium or large. What input and output representation would you use for this problem when applying k-nearest neighbors and explain the advantages and disadvantages of your representations.
  2. How would you represent hypotheses for a genetic algorithm with one or more rules for the above problem. Make sure to precisely define your bit strings, your crossover and mutation operators.
  3. Exercise 10.2, page 303