CS 448 Midterm Review -- Fall 2011

  1. Machine Learning topics
  2. Other ideas
  3. Java
  4. Sample questions
    1. Terms and stuff
      1. What is the definition of intelligence?
      2. What is the definition of learning?
      3. What is the definition of machine?
      4. What is the definition of consciousness?
      5. How does consciousness arise?
      6. How is consciousness like an anthill?
    2. Perceptrons
      1. What is the task of a perception?
      2. For what types of tasks are perceptrons useful?
      3. Could a ptron be used to categorize English sentences as grammatical or ungrammatical?
      4. Describe the perceptron learning rule, both with pseudocode and a formula.
      5. Are the exact values of theta and eta important? Why?
      6. If your ptron has 100 weights, each of which an int, how many possible states can its memory assume?
      7. Are perceptrons good or bad at generalizing? How could you modify a ptron so it would be more likely to be good at generalization?
      8. Why can't a ptron learn XOR? How can you solve this problem?
    3. Genetic algorithms
      1. Write pseudocode for the main loop of a genetic algorithm.
      2. What are some choices you must make in implementing a GA?
      3. How can you measure the performance of a GA (or other adaptive algorthm)?
      4. Is there any relationship between the choices made and the performance of the algorithm? Explain.
      5. Can a GA work without crossover?
      6. Why might a dynamic fitness function be interesting?
      7. Why is convergence a problem?
      8. What's the problem with static exogenous fitness functions?
      9. If the GA is playing a game, and fitness the an individual's score in a round robin tournament with the others in the population, then is the fitness function static or dynamic? Endogenous or exogenous?
      10. When is it important to have large populations?
      11. When are small populations sufficient?
      12. Is there any correspondance between the size of the population and the usefulness of crossover? Explain?
    4. Explain the minimax algorithm.
    5. What is quasimorphism?
    6. You are here (at Willamette) for some reason. Assuming you made a conscious decision to attend college, what prediction did you make that helped with that decision? When will you discover whether that prediction was true? How will you adjust your model if it turns out to be completely incorrect? How will that influence your future behavior?
    7. Was the learning in your pawn program an instance of prediction/correction? Explain.
    8. How should we understand the world (or the brain), by reductionism or wholism?
    9. In your opinion, is the Turing Test a good measure of machine intelligence? Explain.
    10. If someone approached you, attempting to sell an "intelligent" machine, how would you determine whether of not it was intelligent?
    11. In what sense, are organizations like standing waves?
    12. What's wrong with the following argument? "Education is a powerful enforcer of culture. In modern America, children attend school for 6-7 hours a day and are inculcated with a set of ideas. This may be disastrous for our society. Since a person's brain can only contain a limited number of ideas, assimililating one complex of ideas may make it impossible to learn another. We may be crippling our population by teaching them the wrong things.".
    13. What is a meme? How is is like a gene?
    14. If our bodies are simply containers to transport our DNA through time, is consciousness a good thing?
    15. Most religions incorporate notions of good and evil; why?
    16. "In terms of cultural evolution, AIDS will have a powerful effect on fitness."; explain this claim.
    17. Why might it be useful to develop theories about complex adaptive systems in general?
    18. Does the inside of your computer get wet if it simulates a hurricane? Why?
    19. "Trying to create intelligence with a digital computer is like climbing a tree to the moon.". What's wrong with this simile? What's right about it?
    20. Using this definition: "Intelligence - The capacity to acquire and apply knowledge. The faculty of thought and reason.", name 3 intelligent systems (natural or artificial) that do not have emotion. (No fictional characters please!).
    21. Name 3 intelligent systems that do not have consciousness.
    22. If you can't think of any intelligent systems without emotion or consciousness, does that mean there aren't any? Does it mean that there couldn't be one?
    23. How could you use a Hebbian network do to problem solving?
    24. Why is it important to modify Hebb's original learning rule?
    25. Is machine intelligence possible in principle?
    26. Could your mind by transferred into a digital computer? Explain why or why not.
    27. Describe the activity control mechanisms in feed-forward, error back-propagation networks.
    28. Describe Hebb's cell assembly notion.
    29. Why is feedback important in cell assemblies?
  5. Java
    1. Why is it useful to write toString when you first write a class?
    2. Why is it useful to write accessors?
    3. Why does layout in a program matter?
    4. Why are good variable names important?
    5. Why is it important to keep your code simple?
    6. Why is it important to test your code as you write it (instead of writing it all at once)?
    7. Why is it important to separate your conceptualization of the problem to be solved and the code that implements that solution?
    8. If you are programming and find yourself confused, what's the first thing you should ask yourself?
    9. When your code does not work (i.e. has bugs) and you don't know what's wrong, what should you do? Name 5 techniques you might employ to get unstuck.
    10. Write pseudocode for your pawn rote learning.
    11. What would go wrong if you tried similar rote learning for a game like chess?
    12. Write a Population class that creates n individuals with chromosomes of length L.
    13. What is the difference between a deep and a shallow copy?
    14. Why might it be useful to get in the habit of writing toString(), clone() and accessors for every class when you first write them?