Schedule of Classes (as of July 2008)

 

1.   (Aug 25) Lecture 1: Course intro, syllabus, what is an intelligent agent
Homework: Read Asimov_s short story “Runaround”

 

Part I: Fundamentals of Agents and MultiAgent systems.

 

2.   (Aug 27) Lecture 2: Beliefs, desires, intentions (BDI), Satisficing and bounded rationality, Begin BDI logics.

3.   (Sept 02) Lecture 3: BDI Logics Continued, BDI Architectures (e.g. PRS, Soar), Reactive Plans.

4.   (Sept 04) Lecture 4: Decision Theory I; Making simple decisions under uncertainty; risk averseness, risk neutrality; begin Sequential Decisions under uncertainty, Markov Decision Problems (MDPs)

Use Asimov “Runaround”

5.   (Sept 09) Lecture 5: Decision Theory II; MDP Value Iteration; Introduction to Partially Observable Markov Decision Problems (POMDPs)

Homework: Read Asimov’s Chapter I (for history of Science Fiction)

6.   (Sept 11) Lecture 6: (INVITED LECTURE) [EXAMPLE: Prof. Anne Balsamo (School of Cinema) “History of Science Fiction”]

7.   (Sept 16) Lecture 7: Game Theory I: Normal form and extensive form games, Prisoner’s dilemma, Chicken game, Dominance, Iterative dominance, Nash equilibrium, Mixed strategy Nash equilibrium.

“Star Trek: The next generation” episode “The enemy” (Season 3)

8.     (Sept 18) Lecture 8: Game Theory II: Iterative Prisoner’s Dilemma, Stackelberg Games, Bayesian Games, Harsanyi transformation

Need a new science fiction episode for mixed strategies.

9.  (Sept 23) Lecture 9: Agents and emotions; Moral emotions; Behavioral Game Theory intro

    “I, Robot” clip.

10. (Sept 25) Lecture 10: Auctions: First Price, second price (Vickrey auctions); Sequential auctions

11. (Sept. 30) Lecture 11: Agent Learning I: Single agent learning (basics)

Star Trek: The next generation episode: The offspring

Homework: Read Asimov’s “Little lost robot”

 

Part II: Multiagent Interactions

 

12. (Oct 2) Lecture 12: Agent Modeling I: Symbolic plan recognition, model tracing, prediction

13. (Oct 7) Lecture 13: Agent Modeling II: Recursive agent modeling, Plan randomization for adversarial domains

14. (Oct 9) Lecture 14: Biologically inspired multiagent systems, ant algorithms, emergent coordination

15. (Oct 14): Midterm

16. (Oct 16): (Oct 16) Lecture 16: Teamwork I: What is teamwork, team logic, mutual beliefs, joint persistent goals

“Minority Report” clip

17. (Oct 18) Lecture 17: Teamwork II:

a.   Practically implementing teamwork beyond joint persistent goals: representing team plans and roles in an agent architecture, addressing practical communication costs, team monitoring and recovery from failures.

b.   Introduction to decision theoretic approaches to teamwork, distributed POMDPs.

Homework: Vernor Vinge “Fast Times at Fairmont High” from “Hard SF Renaissance”

18. (Oct 23) Lecture 18: Intelligent agents field trip at USC’s Institute of Creative Technology (ICT), in Marina del Rey.

19. (Oct 28) Lecture 19: Team formation (symbolic matching, combinatorial auctions), task allocation (contract nets), coalition formation.

Star Trek episode: Who watches the watchers

Homework: Bruce Sterling “The Swarm”

20. (Oct 30) Lecture 20:   Distributed constraint reasoning, introduction to distributed constraint optimization (DCOP)

21. (Nov 4) Lecture 21: Agent learning II: Multiagent learning. Focus on the multiagent aspect of learning, e.g. in game contexts.

 

Part III: Agents and their impact on society

 

22. (Nov 6) Lecture 22: Show initial clip of commander data goes on trial; provide readings to students in preparation for the trial of commander data. Students divided into two groups, with each group divided into subgroups of 4 each, with each subgroup of 4 given one topic: (i) self-awareness and consciousness; (ii) rights and responsibilities; (iii)…

Star Trek: The next generation “The measure of man”

 

We will show the full episode of trial of commander data (a major part of it). Then we will form teams. We will assign readings. Readings will be photocopied from books or papers. Example readings include:

·      Searle’s “Minds, brains and programs”

·      Nagel’s “What is it like to be a bat?”

·      Turing’s “Computing Machinery and Intelligence” – common counterarguments are presented

·      Roger Penrose “Can a computer have a mind”

23. (Nov 11) Lecture 23: Rights of agents: Students run trials.

Commander data on trial ends.

24. (Nov 13) Lecture 24: Invited Lecture (Melinda Snodgrass, writer “Measure of Man”)

25. (Nov 18) Lecture 25: Adjustable autonomy, Mixed-Initiative Planning, Decision theoretic approaches, strategies in adjustable autonomy

Homework: View one of:

 

“2001: A Space Odyssey” or other movies or TV episodes, where agents or robots are presented in a negative light, and present a design (or a non-technical argument) for safety of agent based systems presented in class in a week’s time.

26. (Nov 20) Lecture 26: New topic: Social networks? Or perhaps preference elicitation from people?

27. (Nov 25) Lecture 27: Safety in agent-based systems: Student presentations. Students should present either a technical design that avoids harm by the robot or AI system they chose to modify; or a non-technical argument as to how this harm could be avoided via societal modifications or new rules and regulations, or explain why this harm will never arise.

28. (Dec 2) Lecture 28:  Agents in the real-world: Massive, DS-1, ARMOR, Mix of applications.

29. (Dec 4) Lecture 29: Wrapup discussion, Review, difference between science fact and science fiction. 

_

 

 

 

Schedule of Assignments and Exams

 

Assignments:

 

 

Midterm: OCT 14, 2008

Final: Finals week