Course syllabus

COMPSCI 767: Intelligent Software Agents Semester 1 2019

 

1 Course description

This course will look at intelligent software agents in multi-agent settings and as an individual intelligent software agent. 

A multi-agent system consists of multiple interacting agents who share the same environment. Such systems are useful for multi-party, complex, real-time problems with uncertainty that are often impossible or hard to model or solve using a single agent alone. Applications of multi-agent systems range from negotiation, cooperating robots,  market and auction analysis, to security. A main theme in this field involves strategic agents where game theory is an important tool. We will be looking at the algorithmic and game-theoretic foundations of multi-agent systems in this course.

One of the core abilities of an intelligent agent is to be able to solve problems. Search is a general purpose technique for finding solutions to problems. However, these search spaces can be quite large and we need to be able to reduce the size of the search space in order to solve problems in a reasonable amount of time and space. We will be exploring several state of the art techniques for reducing the size of these search spaces.

2 Learning outcomes

The students will be able to:

  • General
    • Critically read and evaluate technical papers.
  • Multi-agent systems
    • to understand what a multi-agent system (MAS) is and when they are useful
    • to be able to explain important challenges of MAS such as coordination and cooperation
    • to be able to apply some well-known distributed optimization algorithms
    • to phrase MAS scenarios using the language of game theory
    • to be able to identify or derive equilibria in normal form or extensive form games
  • Intelligent software agent
    • to understand current research papers in Heuristic Search
    • to present problems in PDDL, the standard language for describing planning problems
    • to understand some of the tradeoffs involved in using current
      techniques to reduce the problem space sizes
    • to create domain-specific heuristics 
    • to understand some of the tradeoffs involved in creating heuristics

 

3 Teaching staff

Mike Barley (Coordinator)

Room: 488, Computer Science Building (Building 303S)

Phone: 373-7599, Ext 86133

Email: barley@cs.auckland.ac.nz

 

Jiamou Liu

Room 487, Computer Science Building (Building 303S)

Phone 373-7599, Ext 89528

 Email: Jiamou.liu@cs.auckland.ac.nz

 

Class Rep:
Alexander Swain
Email: aswa408@aucklanduni.ac.nz

 

4 Lecture Times

Tues 10am Room 302-G20

Wed 12pm Arts 1, Room 203  

Thu 2pm Room 302-G20

 

5 Assessments

Your final grade will consist of a number of internal marks worth 50% combined and an exam worth 50%. The internal assessment for Jiamou will be 25%, including:

  • an assignment worth 15%
  • reading worth 10%

The internal assessment for Mike will also be 25%, including:

  • an assignment worth 10%
  • 15 quizzes worth 15% in total

 

5.1 Internal Marks

Jiamou’s assignment will be available on 8 April and due via the dropbox on 22 April by 11:59pm

Jiamou’s reading list will be available in Week 2 of the semester.

Mike’s assignment will be available on 23 April and due via the dropbox on 24 May

Mike’s quizzes: there will be 15 quizzes, a quiz for each of the last 15 classes.

 

6 Proposed lecture schedules (subject to change)

Week 1 Introduction to Multi-agent Systems

Week 2 Distributed Optimisation

Week 3 Introduction to Game Theory

Week 4 Social Structures

Week 5 Reinforcement Learning

Week 6 Research Paper Reading and Presentations

Week 7 Review of Search Techniques

Week 8-12 Research Paper Reading and Presentations (See link below  to access papers and their dates)

 

6.1 Jiamou's Part of the Course

Jiamou will teach the first half of the course and Mike will teach the second half. 

In Jiamou’s half of the course, the lectures will focus on the algorithmic, game-theoretic, graph-theoretic, and learning of agents in a multi-agent system. After five weeks’ of lectures, there will be a reading task where the students will form groups, read and present a recent research article. The list of papers will be posted in week 2 of the semester. Each paper will be taken from an AI conference and will be 7-8 pages long. Jiamou’s part of the exam will ask questions about the lecture contents.

 

6.2 Mike's Part of the Course

6.2.1  Overview

In Mike’s half of the course, after the review of search techniques, there will be a paper assigned to be read for each class.  The list of papers will be posted before the mid-term break.  Each paper will be taken from an AI conference and will be 10 pages or less.  You will be expected to read the paper and to understand the main points of the paper before the class.  A class will have three parts:

  1. There will be a marked quiz on the paper at the beginning of class on the assigned paper.
  2. The class will break into groups to discuss the paper and compare/contrast it with the other papers that have been read in the class.
  3. The groups will report back to the class about their discussions.

Mike’s part of the exam will primarily ask questions about material found in these papers.  All-in-all, 40% of your grade will be based on your understanding of these papers!!!

6.2.2 Reading List

Here is a list of the papers you will need to read for my part of the course including the date on which there will be a quiz and discussion on that paper.  Here are examples of the types of questions that may be asked on the quizzes and discussed during class.

The papers are technical conference papers of less than 10 pages.  You will probably need to spend 1.5 hours in reading and analysing a paper.  That is 4.5 hours per week just on the papers.  I believe the best way to do this is through group discussions before class, either in person or via Piazza or something similar.  That will be a lot more enjoyable and productive than just doing it on your own.  You will be quizzed on each paper which adds up to 14% of your grade and if you do all 14 quizzes (regardless of your marks on those quizzes) you will get an extra mark, making it 15% of your grade.

 

6.2.3 Reading Groups

The class will divide up into reading groups of ~3 people.  Each class period the class will break up into their individual reading groups and be assigned discussion questions.  The groups will discuss their questions and then come back together to tell the class the results of their discussions.  The reading groups have names and this is a link to   the list of reading groups that I know about.

 

7 Seeking assistance

The primary source of assistance is the teaching staff. Please contact Mike or Jiamou with any questions or concerns about the course. Both are available via email.

 

For help with more generic study skills or literacy, the Student Learning Centre and Library both offer many courses designed to help students become more efficient at study.

 

7.1 Missed lectures

Overhead slides and recommended reading will be provided. Please review the material prior to seeing the teaching staff. If you know in advance that you will miss a lecture, please let the course coordinator know.

 

7.2 Exam

The final exam is worth 50% of your final mark. Please check Student Services Online for the exam time and date. The exam is closed book, calculators are not permitted. Provisional exam results can be obtained from Student Services Online.

 

7.3 Missed exam

If you miss the exam for any valid reason, or you sit the exam but believe that your performance was impaired for some reason, then you may be able to apply for an aegrotat, compassionate or special pass consideration. For more detailed information, refer to pages 44–45 of the University of Auckland’s 2012 Calendar.

 

7.4 Policy on Cheating and Plagiarism

Cheating is viewed as a serious offence by the University of Auckland. Penalties are administered by the Discipline Committee of the Senate, and may include suspension or expulsion from the university. Do not copy anyone else’s work, or allow anyone else to copy from you.

 

For more information on the University’s policy on cheating, please refer to the web page: http: //www.auckland.ac.nz/uoa/home/about/teaching-learning/honesty

Course summary:

Date Details Due