Artificial Intelligence 4 - homepage

This page will hold day-to-day information about the AI4 module, taught primarily by Roderick Murray-Smith (RMS)

The moodle page for this course is http://fims.moodle.gla.ac.uk/course/view.php?id=145 but it will be less functional than this page

NEWS

Hand out October 19th, hand-in November 25th (9am). You should hand in your printed report to the appropriate box. You should also use e-mail to e-mail the python source code for your solutions to rod@dcs.gla.ac.uk. Use the subject AI4-AE in your e-mail.
This assignment will have three parts. The practical parts are based on the Berkeley Pac-Man exercises, which should give you experience at implementing a number of the topics you have learned about so far. The third part gives you some freedom to develop in your own direction, and push your programming and AI skills as far as you can. As mentioned in the lecture, ou can use the report to discuss the rationale behind your implementation of the first two parts of the project.
1. Complete the Search in Pac-Man assignment http://inst.eecs.berkeley.edu/~cs188/pacman/projects/search/search.html You will fill in portions of search.py and searchAgents.py during the assignment. You should submit these two files along with your report. (30%)

clip_image004.jpg 2. Complete the Multiagent Pac-Man project http://inst.eecs.berkeley.edu/~cs188/pacman/projects/multiagent/multiagentProject.html You will fill in portions of multiAgents.py during the assignment. You should submit this file with your code and comments. You may also submit supporting files (like search.py, etc.) that you use in your code. (30%)

3. Use of evolutionary approaches to improve the performance of agents in the Pac-Man environment. Use these, in any way you see fit, with the Pac-man environment resources (this can involve techniques from some of the more advanced labs on the Pac-Man site). Describe how you used these methods, provide code, and results of simulations and document these clearly in your report. Include a 4-page literature review on the use of artificial intelligence in the games and entertainment industry, discussing techniques used, citing academic literature. (40%)

Marking scheme

The marking scheme for parts 1 and 2 will follow the credit marks specified in the Berkeley assignments, and will sum to 60% of the assignment.
The report on the use of evolutionary approaches in game agents is worth 60% of the assignment total. The report should be no more than 10 pages of text. An `A' grade report would be well written, citing appropriate literature, show creative use of the AI techniques to solve important and interesting problems in game agent design, and take account of practical constraints. A `B' grade might be due to poorer writing, less insightful use of algorithms, poorer presentation of the simulation results. A `C' grade would have some basic use of evolutionary techniques, but would have shortcomings in the implementation, testing, presentation or justification of choices. A `D' grade report would have the most basic use of evolutionary techniques in the multiagent example, but they would not be well justified, and might not be desirable.

Lecture times

Note PDFs of lectures can only be downloaded from University machines (or while VPNed in).

Lecture	Topic	Readings	Exercises/Labwork/Handouts/WWW Links
1, 27th Sept	Introduction RMS	Ch1	Q1.1,1.2 Computing Machinery and Intelligence, A. M. Turing Turing Page Turing Archive AI's greatest trends and controversies.
2, 28th Sept	Intelligent Agents JHW	Ch2	Q2.5,2.6
3, 29th Sept	Uninformed search & representation RMS	Ch3	Q3.2,3.7,3.13
4th Oct	Social Signal Processing Understanding Social Interactions through Nonverbal Communication Analysis AV		video (computer frustration)
6th Oct	Tutorial: discussion of DARPA robotics grand challenges. JHW		Stanford racing team videos Stanley video Collection of Stanley videos and papers
4, 11th Oct	Informed search, optimisation and evolutionary approaches Optimisation, design and evolutionary approaches RMS	Ch4	Evolved creatures video (local copy) Evolved creatures - Karl Sims GA Archives Evolutionary Computing Artificial Life Dawkin's Viruses of the Mind Q4.2,4.3
5. 12th Oct	Game playing RMS	Ch6	Play Chinook on the web. Jonathan Schaeffer. Man Versus Machine: The Silicon Graphics World Checkers Championship , Technical Report 92-19, Department of Computing Science, University of Alberta, 1992. Jonathan Schaeffer, Joseph Culberson, Norman Treloar, Brent Knight, Paul Lu and Duane Szafron. A World Championship Caliber Checkers Program , Artificial Intelligence, vol. 53, no. 2-3, pp. 273-290, 1992. Jonathan Schaeffer, Robert Lake, Paul Lu and Martin Bryant. Chinook: The World Man-Machine Checkers Champion , AI Magazine, vol. 17, no. 1, pp. 21-29, 1996. Jonathan Schaeffer and Robert Lake. Solving the Game of Checkers , Games of No Chance, Richard J. Nowakowski (editor), Cambridge University Press, pp. 119-133, 1996. Herbert A. Simon and Jonathan Schaeffer. The Game of Chess , Handbook of Game Theory with Economic Applications, vol. 1, Robert J. Aumann and Sergiu Hart (editors), Elsevier Science Publishers, Netherlands, pp. 1-17, 1992. Also available as CMU technical report AIP-105. A chess-playing machine, Claude E. Shannon QQ6.1,6.2,6.3,6.4,6.16
6. 13th Oct	Uncertainty RMS	Ch 13	Q13.8,13.10,13.15
7. 18th Oct	Decision making under uncertainty RMS	Ch 16	Tutorial Introduction to Decision Theory, D. Warner North Judgment under Uncertainty: Heuristics and Biases, Amos Tversky & Daniel Kahneman Q16.2,16.11 Try Qs 16.1, 16.4 with classmates
8. 19th Oct	Intro to Belief Nets & Expert systems issues RMS	Ch 14.1-14.3,14.7 (no detailed questions on inference algorithms)	Q14.1,14.3
20th Oct	Getting going with the Berkeley Pac-Man environment (in Boyd Orr lab) RMS		Complete the Search in Pac-Man assignment http://inst.eecs.berkeley.edu/~cs188/pacman/projects/search/search.html
9. 25th Oct	Robotics RMS	Ch25	A Robust Layered Control System for a Mobile Robot, Rod Brooks Outline for a theory of Intelligence, Albus Stanford racing team videos iRobot corporation, research Hans Moravec's home page Q25.6,25.8
10. 26th Oct	Perception RMS	Read Ch24, focus on Ch24.1,24.6,24.7	Sine-wave speech demo McGurk effect Gregory face videos
	Support for assessed exercise RMS
11. 1st Nov	Machine Learning RMS	Ch 18.1-18.3,18.6	Q18.2,18.3,18.4
12. 2nd Nov	Machine Learning, Self-organisation. RMS
12. 8th Nov	Neural networks RMS	Ch20.5	Q20.11,20.19
13. 9th & 15th Nov	Neural networks, cont RMS	Ch20.5	Q20.11,20.19
15. 16th & 22nd Nov	Affective Computing RMS		TR#515: Toward Agents that Recognize Emotion Rosalind W. Picard Microsoft Adaptive Systems & Interaction group
17th Nov	NN lab RMS		Example MATLAB programs. Run mlpdemo, SOMdemo, SOM3Ddemo within MATLAB, explore the behaviour with your own classification challenges, and try changing the number of hidden units in the model, or adjust the regression problem and noise levels). Old Lab on neural nets
23rd Nov	No lecture - free for work on assessed exercise RMS
17. 29th Nov	Ethical Issues in AI RMS
18. 30th Nov	Philosophical Issues RMS	Ch26,27	Q26.1,26.7 Computing Machinery and Intelligence, A. M. Turing Turing Page Turing Archive

Course Textbook

Artificial Intelligence: A Modern Approach, 3rd Edition,
by Stuart Russell and Peter Norvig.
2010
ISBN: 978-0132071482

Required text - contents of chapters specified in lectures will be examinable.

Book home page - http://aima.cs.berkeley.edu/

Some Other Resources

AI Topics - a library of information on AI - really useful set of reasonably recent articles and overviews.
Russell's AI directory
Lecture Slides from Russell's Book Homepage (note the pdf files don't seem to work too well - .ps more reliable)

Other points

Selected questions for each lecture/chapter. These are to give an indication of the sort of level of knowledge needed in the exam, and to highlight significant points from the course. Worked answers were handed out in class. some more worked examples can be seen in the revision lecture.
SAMPLE EXAM PAPER
Exam paper 2000-2001 (RTF format) pdf format
Exam paper 2001-2002 (pdf)
Exam paper 2005 Sample Answers
No exam papers for 2006, 2007 (AI4 didn't run those years).