Evolutionary Computation

CSE 584, Winter Quarter 2003


Time : Mondays and Wednesdays, 4:00-5:20pm
Location Room 110A in the 1600 building.

Instructor: Melanie Mitchell (mm@cse.ogi.edu)
Department of Computer Science and Engineering
OGI School of Science & Engineering
Oregon Health & Science University

Office : CSE Central 145
Telephone : (503) 748-1455
Office hours : Mondays and Wednesdays, 2:00-4:00pm, and by appointment.

Prerequisites: Programming experience in any high-level language.

Text: An Introduction to Genetic Algorithms by Melanie Mitchell.
MIT Press, 1996. ISBN: 0-262-63185-7.

You can find solutions to selected thought exercises from this book in pdf format here.

Assignments: There will be weekly homework assignments, both written problem-solving and computing, during the first half of the quarter. The second half will be devoted to completing a larger project involving evolutionary computation. The project will consist of either replicating (with variations) a research project in the literature, or an original (tractable) research project. By the sixth week (February 10), students will have written a one-to-two page project proposal, and the remaining four weeks will be devoted to completing this project. By March 10, each student will complete a first draft of a paper describing their project, in the style of a conference paper, 8-10 pages. Each of these will be reviewed (in the style of a conference paper review) by three other students in the class. The reviews will be returned to each student by March 12. The students will then revise their papers according to the reviews. Each student will also make a 15-20 minute presentation to the class on their research, with an additional five minutes for questions from the audience. The final papers will be due March 21.

Grading: 50% of the grade will be based on the five homework assignments during the first half of the quarter. The other 50% will be based on the project, including the project proposal, the in-class presentation, the final paper, and the reviews of other students' papers.

More specifically, each of the five homework assignments will be worth 100 points: 40 for the written exercises and 60 for the computer exercises. The maximum number of points for the homework is thus 500. The project will also be given a maximum of 500 points, distributed as follows: Project proposal (50), project writeup (300), reviews of other students papers (50), project presentation (100).

Academic integrity: The students will be responsible for following the OGI guidelines for academic integrity. You may discuss the general concepts and principles behind an assignment with other students. In fact, you are encouraged to do this whenever possible, because it is often a valuable way to reinforce ideas, and to learn new perspectives. However, in doing assignments, each student is expected to develop, write up, and hand in an individual solution and, in doing so, develop a sufficient understanding of the problem and solution so as to be able to explain it adequately to the instructor. Under no circumstances should a student copy or consult the completed solution of another student.

Syllabus (subject to change):

Date

Topics

Reading
(to be done before lecture)

Homework

Monday January 6

Introduction to evolutionary computation; a simple genetic algorithm in C.

None

HW 1 assigned; due Monday January 13.

Wednesday January 8

Introduction to evolutionary computation; a simple genetic algorithm in C.

Required reading: Textbook, Chapter 1; handout of C code.


...

Monday January 13

Evolving cellular automata; coevolutionary learning

Required reading: Textbook, Chapter 2, Section 2.1; paper by Krzysztof Krawiec (handed out in class).

Optional reading: Paper by Mitchell, pdf format ; paper by John Koza postscript format.

HW 1 due.

HW 2 assigned; due Wednesday Jan 22.

Wednesday January 15

Coevolutionary learning, continued; genetic programming

Optional reading: Paper by Pagie and Mitchell, pdf format ; Paper by Shapiro, "Does Data-Model Co-evolution Improve Generalization Performance of Evolving Learners?" (retrieve gzipped postscript from this site ; Paper by Rosin and Belew, postscript format .

...

Monday January 20

Martin Luther King day: OGI holiday.

None

...

Wednesday January 22

Evolutionary computation applied to data analysis and forecasting.

Required reading: Textbook, Chapter 2, Section 2.2 .

Optional reading: Paper by Allen and Karjalainen: pdf version .

HW 2 due.

HW 3 assigned; due Monday January 27.

Monday January 27

Evolving neural networks.

Required reading: Textbook, Chapter 2, Section 2.3.

Optional reading: Paper by Stanley and Miikkulainen, pdf format .

HW 3 due.

HW 4 assigned; due Monday February 3.

Wednesday January 29

Genetic algorithms for image processing; evolutionary art

Required reading: (1) Paper by Harvey et al.: pdf format ; (2) Karl Sims' web site on "Genetic Images", including review by Kevin Kelly and review by George Fifield, both linked to that site.

Optional reading: (1) Paper by Sims: html version ; (2) Paper by Johnson et al., "Evolving Visual Routines", which you can download by clicking here .

...

Monday February 3

Evolutionary robotics

Required reading: (1) Paper by Karl Sims from Siggraph 1994 , gzipped postscript format ; (2) paper by Lipson and Pollak from Nature , pdf format ; (3) commentary by Brooks, pdf format .

Optional reading: (1) Paper by Karl Sims from Artificial Life , gzipped postscript format .

HW 4 due.

HW 5 assigned; due Monday February 10.

Wednesday February 5

Implementation issues for genetic algorithms.

Textbook, Chapter 5

...

Monday February 10

Self-adaptation

Required reading: Paper by Spears, gzipped postscript.

Optional reading: Paper by Bedau and Seymour, gzipped postscript ; paper by Altenberg, pdf.

HW 5 due.

Project proposals discussed this week. 1-2 page written version of proposal due Wednesday, February 19. Individual meetings with professor, to be scheduled.

Wednesday February 12

Modeling with evolutionary algorithms: Baldwin effect and evolutionary reinforcement learning.

Required reading: Textbook, Chapter 3, Section 3.1

...

Monday February 17

Presidents' day: OGI holiday

None

...

Wednesday February 19

Modeling with evolutionary algorithms: Sexual selection, signaling, social behavior

Required reading: Textbook, Chapter 3, Section 3.2.

Project proposal due.


Work on project.

Monday February 24

Modeling with evolutionary algorithms: artificial immune systems; evolutionary aspects of cancer

Required reading: Paper by Forrest and Hofmeyr, postscript format or pdf format .

Optional reading: Paper by Maley and Forrest, postscript format .

...

Wednesday February 26

Artificial stock market with evolving agents

Required reading: Paper by Joshi and Bedau, postscript format .

Optional reading: Paper by Palmer et al., pdf format .

...

Monday March 3

Theory of evolutionary computation, part 1

Required reading: Textbook, Chapter 4, Section 4.1.

Work on first draft of project writeup; due Monday March 10.

Tuesday March 4

(note day change for this class)

Theory of evolutionary computation, part 2; Prospects for evolutionary computation

Required reading: Textbook, Chapter 4, Sections 4.3--4.4. Optional reading: Textbook, Chapter 6

...

Monday March 10

Student project presentations

None

First draft of project writeup due: hand in four copies.

Reviews assigned, due Monday March 17.

Wednesday March 12

Student project presentations

Student papers.

...

Monday March 17

Student project presentations

None

Reviews due (please hand in two copies of each review). Work on revisions of project writeup.

Wednesday March 19

Student project presentations

None

Final project write-up, due March 21.