Artificial Intelligence: Assignment 7 Solutions

Problem 1 (4 points): Read Section 20.5 (Neural Networks) of the textbook. In your own words describe the important differences between neural networks and decision trees as approaches to learning. For each give an example of a real-world problem that is suited to that style of learning.
Justify your answers.

The following table outlines the important differences and provides examples:

AttributeDecision TreesNeural Networks
Problem solvingAlgorithmicAdaptive
InputCategorical data*Continuous real numbers
OutputDiscrete classes with certainty and explanationsDiscrete classes
Solutions correspond toFOL conjunctionsArbitrary mathematical functions
Robustness to noiseLimitedVery good
ApplicationsCredit risk analysisAutonomous driving
Marketing (Consumer behavior prediction)Face recognition

*Note: Decision Trees can be adapted to deal with continuous real numbers as input, but their capabilities are very limited when working with these kinds of attributes.

Problem 2 (6 points):

For full credit on both assignments your program had to build a decision tree from the following training sets and then classify the corresponding test sets.

Coins Example: Training, Test

Books Example: Training, Test

Weather Example: Training, Test

Food Example: Training, Test

The solution to all the examples can be found at the updated page here.