School of CSE - Details Needed for New Course Proposal

Course Title: Knowledge Representation and Reasoning

Proposers: Prof. Norman Foo, Dr. Thomas Meyer, Dr. Maurice Pagnucco, Dr. Jochen Renz

Rationale
Why is the new course being proposed?

Knowledge Representation and Reasoning (KRR) is at the core of Artificial Intelligence. All facets of AI make use of KRR to some extent or other. With the presence of NICTA and its Knowledge Representation and Reasoning Program we have a large body of expertise in this area. Further NICTA Level E appointments in the area of constraint programming will further bolster our expertise and resources. Existing courses on AI cover KRR on a very basic level, by mainly introducing propositional and first-order logic and by presenting some historical KRR concepts like rules, semantic nets and frames. We want to give students an advanced and more up-to-date introduction to KRR. This covers mainly recent trends and current research issues with which our group has expertise.

What are the academic objectives?

Introduce students to current research issues in KRR and prepare them for a thesis or project.

Which programs/stage does it serve?

BSc, BSE, BCE 4th year; MSc/PhD

Why can the same objectives not be achieved with existing courses?

Current AI courses cover KRR only as one of many topics and give only a very basic introduction and historical overview.

How does the proposed course relate to other courses?

It is an extension of the AI courses COMP3411: Artificial Intelligence, COMP9414: Artificial Intelligence and is directed towards students who are interested in AI and who want to that want to specialise. It is related to the courses COMP4415: Logical Foundations of Artificial Intelligence, COMP9417: Machine Learning, COMP4411: Experimental Robotics, COMP4416: Intelligent Agents and to some extent COMP9444: Neural Networks. It complements them all quite nicely.

What overlap is there?

The AI courses COMP3411, COMP4415, COMP9414 give a basic introduction to KRR. We build on that introduction and deal with more advanced topics. In the first three hours we give a short summary of these basics, which is the only overlap.

If there is any overlap, why is this justified/not a problem?

The overlap consists only of a short recap of what was done in the introductory AI courses regarding KRR.

Stakeholders and Consultation

Who are the potential stakeholders, who was consulted about the proposal (inside the School as well as outside), what was the result of that consultation?

The main consultation was held with the Artificial Intelligence researchers within the department.

  • Mike Bain
  • Alan Blair
  • Paul Compton
  • Achim Hoffman
  • Eric Martin
  • Bill Wilson
  • Claude Sammut
  • Arcot Sowmya
  • Ron van der Meyden

    Enrolment Impacts

    Likely enrolment (with justification), and impact on enrolments of other courses.

    We expect that a fourth to a third of the attendants of the three offered AI courses attend this course, so this course should have the same size as the introductury AI courses. It might impact other advanced AI courses such as COMP9417 Machine Learning, COMP9441 Cryptography and Security, COMP9444 Neural Networks, COMP9511 Human-Computer Interaction, COMP9517 Computer Vision, COMP9518 Pattern Recognition and Vision in that students who want to attend further AI courses decide to take our course instead of one of the others.

    Justification of Prerequisites (or lack thereof)

    Prerequisites:
    At least 12CP in COMP3xxx courses or above including one of the introductory AI courses COMP3411, COMP4415, COMP9414

    Any Courses this is Replacing, and Why?

    None.

    Delivery and Assessment

    3 hours per week.
    4 assignments (15% each)
    1 project, can be done in pairs (40%)
    no exam

    Anything noteworthy about delivery mode, assessment (with justification).

    Handbook Entry

    Knowledge Representation and Reasoning (KRR) is at the core of Artificial Intelligence. It is concerned with the representation of knowledge in symbolic form and the use of this knowledge for reasoning. This course presents current trends and research issues in Knowledge Representation and Reasoning (KRR).
    It enables students interested in Artificial Intelligence to deepen their knowledge in this important area and gives them a solid background for doing their own work/research in this area. The topics covered in more detail are AI Logics, Probablilistic Reasoning, Constraints, and Game Theory. We require that participants have completed at least one introductory AI course (COMP3411, COMP4415, COMP9414).

    Textbooks/References

    Textbook: none
    References (per topic):

    1. Logic/ 2. AI Logics:

    (a) Ronald J. Brachman and Hector J. Levesque. Knowledge Representation and Reasoning, Morgan Kaufmann, 2004. (http://books.elsevier.com/us//mk/us/subindex.asp?maintarget=&isbn=&country=United+States&srccode=&ref=&subcode=&head=&pdf=&basiccode=&txtSearch=&SearchField=&operator=&order=&community=mk)

    (b) Michael R. Genesereth and Nils J. Nilsson. Logical Foundations of Artificial Intelligence, Morgan Kaufmann, 1987.

    3. Probabilistic reasoning:

    (a) Judea Pearl, Probabilistic Reasoning in Intelligent Systems: Networks of Plausible Inference, Morgan Kaufmann, 1988.

    (b) Stuart Russell and Peter Norvig, Artificial Intelligence: A Modern Approach, Prentice Hall, 2nd Edition, 2003.(http://aima.cs.berkeley.edu/)

    4. Constraints:

    (a) Rina Dechter. Constraint Processing. Morgan Kaufmann, 2003.

    (b) Grzegorz Kondrak and Peter van Beek. A theoretical evaluation of selected backtracking algorithms. Artificial Intelligence, 89:365-387, 1997. (http://ai.uwaterloo.ca/~vanbeek/publications/ai97.ps.gz)

    (c) A. G. Cohn and S. M. Hazarika. Qualitative Spatial Representation and Reasoning: An Overview. Fundamenta Informaticae, 46 (1-2), pp 1-29, (2001). (http://www.comp.leeds.ac.uk/qsr/pub/funinfreview.ps)

    (d) J. Renz, Qualitative Spatial Reasoning with Topological Information, LNCS 2293, Springer, 2002. (http://www.springer.de/cgi-bin/search_book.pl?isbn=3-540-43346-5)

    5. Game theory:

    (a) Selected lecture slides from the site http://gametheory.net/html/lectures.html.

    (b) Luce and Raiffa: Games and Decisions. Dover Publications.

    Syllabus

    Indicative syllabus / overview of contents (at a level of detail well beyond that of the handbook entry)

    1. Logic Recap (1 week)
    2. AI Logics (3 weeks): Non-monotonic Reasoning, Default Reasoning, Belief Revision,...
    3. Probablilistic Reasoning (3 weeks): Bayesian Nets, POMDP's,...
    4. Constraints (3 weeks): Constraint solving, CLP, Spatial and Temporal Reasoning,...
    5. Game Theory (3 weeks): Introduction, ...
    6. Project presentations (1 week): course participants present their projects. In addition, the Ph.D. students of our group will give a short presentation of their Ph.D. projects

    Effect on School Resources:

    1. Who is proposed to teach the proposed new course, and what impact would this have on their planned/current allocation?

    The teaching will be shared by the four proposers in equal shares. Dr. Meyer and Dr. Renz are NICTA employee. Prof. Foo and Dr. Pagnucco.

    2. What sort of tutorial component is proposed, if any?

    None.

    3. What is the likely impact on lab utilisation (this relates to assignment and project work as well as scheduled labs?

    Only for preparing assignments and the project. Very little impact if any.

    4. Any other resource needs? E.g. special print/disk quota, access to servers, access to special machines, special labs.

    None.