CS 5970 Homework 1 — Evolutionary Algorithms as Problem Solvers

NOTE: The hardcopies of the parts of this assignment are due at the beginning of the class period. This means that if you are even a minute late, you lose 20%. If you are worried about potentially being late, turn in your assignments ahead of time. Do this by submitting them to me during office hours or by sliding them under my office door. Electronic copies are due by 4:00 pm on the due date. Submit them through D2L before the time they are due. Do not send assignments to me through email or leave them in my departmental mail box.

1. Motivation

To use evolutionary algorithms as problem solvers, we need to apply them to particular problems. To do that, we need to (following De Jong, p 72):

• decide what an individual in the population represents,
• provide a means for computing the fitness of an individual,
• decide how children (new search points) are generated from parents (current search points),
• specify population sizes and dynamics,
• define termination criteria for stopping the evolutionary process, and
• return an answer

2. Goals

The goals of this assignment are:

• to give you experience with reading papers from the primary, peer-reviewed literature in evolutionary computation, and
• to give you experience with dissecting such papers and understanding how evolutionary computation models are implemented for particular problems.

3. Assignment

Read the paper "Memetic Learning: A Novel Learning Method for Multi-Robot Systems" by Hougen, Carmer, and Woehrer (from the International Workshop on Multi-Robot Systems, March 2003) then answer the questions below regarding the GA in this paper. You may skim the sections on reinforcement learning with eligibility traces (§3.1.1) and memetic learning algorithms (§3.1.3) to find relevant information without working to deeply understand reinforcement learning or memetic learning. Be sure to pay close attention to the section on genetic algorithms (§3.1.2) where the GA is described.

1. What does an individual in the population represent?
2. Is this a fixed-length linear object, a fixed-length nonlinear object, a variable-length linear object, or a nonlinear variable-length object?
3. How is each individual encoded?
4. Is this encoding genotypical or phenotypical?
5. How is fitness calculated for each individual in the population?
6. Does this method of fitness calculation conform to De Jong’s discussion of this issue (p 77)?
7. What reproductive operators are used?
8. What parameters are used for each of the reproductive operators?
9. What is the parent population size?
10. What is the offspring population size?
11. Is an overlapping or non-overlapping generation model used?
12. What stopping criteria are used?
13. "[W]hat guarantee does one have regarding the properties of the solutions found?" (De Jong, p 79)
14. What answers are returned?

4. What to Turn In

You will turn in both a typed hard copy and a machine readable electronic copy of your homework that answers the questions above.