Mini-projects

Topics

Start with the world of Homework 4 (reinforcement learning in diskoids, which have simple feelers and which need to eat plasmoids).

Some ideas for topics

• Add another type of sensor (audition, sonar, olfaction, vision).
• Make learning more sophisticated (e.g., use a neural network instead of a lookup table).
• Implement a genetic algorithm, and compare it to reinforcement learning or combine it with reinforcement learning.
• Implement learning in separate species that are in competition.
• Allow diskoids to pass information among themselves.

Submission

Submit (in Oncourse Assignments2) your program, your written report, and your oral presentation slides. Only one person in each team needs to make the submission.

Reports

Your oral and written reports should include the following:

1. (Briefly) how the program fits into the big picture (intelligence, adaptation, perception and action)
2. What the program does
3. How the program works, including how to run it if this is not obvious
4. What you learned

Written reports don't need to be longer than two or three single-spaced pages.

Program

Your code should be documented and should say, for each class or function, which team member wrote it.

Be sure your program runs without errors before you submit it.

© 2012. Indiana University and Michael Gasser. www.indiana.edu/~q320/projects.html 2012-04-01