However, an even more dramatic event in the career of [Samuel’s checkers player] had slipped by unnoticed several years before. On some forgotten day, the program had begun to beat Samuel himself; and that, to my mind, was one of the most remarkable events in the history of computing. — J. Copeland, Artificial Intelligence, 1993.

Information Theory

Compute the following using the formulations from Information Theory discussed in class (AIMA, Section 18.3).

Exercise 6.1

Compute the information gain provided by the “Hungry?“ question from the restaurant domain. Use this restaurant data (AIMA, Figure 18.3). Would this question be better than either of the questions discussed in class (i.e., “Patrons?”, “Type?”?

Record your answers in lab06_1.txt (or you can use MD format).

Problem Framing

Work in teams of two (or perhaps three) for this one. Select an ML-compatible problem and work through these Google Crash Course exercises. If you already answered these questions when you worked this unit’s guide, go through them with your partner(s).

Exercise 6.2

Do the Try It Yourself exercises from the Problem Framing course.

  1. Try It Yourself: Framing
  2. Try It Yourself: Formulating

Record your answers to the ten exercises in lab06_2.txt. Each partner should record their own answers and submit them separately, but feel free to share your answers. Please indicate all partners’ names in your submission.

NumPy

Do the following using this class example as a guide.

Exercise 6.3

Do the following things using NumPy:

  1. Load the Keras version of the Boston Housing Price dataset (boston_housing) and do the following:
    1. Print the number of training and testing examples.
    2. Print the rank (i.e., number of axes/dimensions), shape and data type of the examples.

Save your code in lab06_3.py, including all partner’s names in the file.

Checking in

We will grade your work according to the following criteria:

See the policies page for lab due-dates and times.