Meeting Agenda

Wednesday, August 27^th, 1997

Location:

DC 1304

Time:

13:30

Chair:

Patrick Gilhuly

1. Adoption of the Agenda - additions or deletions

2. Coffee Hour

Coffee hour this week:

???

Coffee hour next week:

???

3. Next meeting

Date:

September 3^rd, 1997

Location:

DC 1304

Time:

13:30

Chair:

Michael McCool

Technical presentation:

Patrick Gilhuly

4. Forthcoming

1. Peter Harwood (9/10)
2. Celine Latulipe (9/17)

1. Michael McCool (9/10)
2. Peter Harwood (9/17)

5. Technical Presentation

Presenter:

Glenn Evans

Title:

Bidirectional Path Tracing

Abstract:

The field of solving global illumination problems is an active field of research. In this talk, I will present recent mathematical formulations that provide a theoretical framework for all global illumination algorithms. There are two approaches taken: discrete approximations such as radiosity and statistical approximations that combine Monte Carlo integration with ray tracing. The focus will be on showing how ray tracers, light tracers and bidirectional path tracers are all simply special cases of a more general algorithms.

6. General Discussion Items

• Participation in frosh grad student tours.

  This was in the mailbox:

  I'm organizing the research group presentations for this year's
  orientation, and we would like each research group to give a 5-min
  presentation to the new students about the research activities in
  their group. The talk will be on Thursday, Sept.4, sometime between
  1pm-~2pm.

  Also, right after these talks from about 2pm-3:30pm, we are hoping to
  have a casual "open house" so new students can talk to researchers in
  various labs. No tours or anything structured is scheduled, we are
  just going to tell students to drop by any of the labs/groups that are
  participating. Please let me know if the CGL would like to do this.

7. Action List

• Net Bee '97

8. Director's Meeting

9. Seminars

        ICR SHORT COURSES:
        
        


        DEPARTMENT OF COMPUTER SCIENCE
        UNIVERSITY OF WATERLOO
        SEMINAR ACTIVITIES


        Master's Thesis Presentation


                       - Wednesday, September 3, 1997


        Russell  Mok,  Graduate Student, Department of Computer
        Science,   University   of   Waterloo,  will  speak  on
        ``Abnormal event handling mechanisms''.

        ROOM:          Davis Centre Room DC1304

        TIME:          2:30 - 3:30 p.m.


        ABSTRACT

        An  abnormal  event  handling  mechanisms  (AEHM)  is a
        crucial   programming   feature  for  building  robust,
        reliable  software,  giving  programmers the ability to
        deal  with  a wide variety of dynamic events.  Examples
        of  abnormal  events are UNIX signals and exceptions in
        Ada and C++.

        While  many  modern  concurrent  programming  languages
        provide  some form of AEHM, the interaction between the
        AEHM  and the concurrency features is weak; in essence,
        the AEHM is only applicable for sequential programming.

        I  will  present  a framework for concurrent AEHM's and
        explain  why  existing  AEHM's  are inadequate.  I will
        show  a  design  for  an  AEHM,  which  addresses  many
        shortcomings    of    existing    AEHM's,    and   it's
        implementation in uC++, a concurrent extension for C++.



         DEPARTMENT OF COMPUTER SCIENCE
         UNIVERSITY OF WATERLOO
         SEMINAR ACTIVITIES


         Master's Presentation


                        - Friday, September 5, 1997


         Vlado  Keselj, Graduate Student, Department of Computer
         Science,   University   of   Waterloo,  will  speak  on
         ``Multi-Agent    Systems   for   Internet   Information
         Retrieval Using Natural Language Processing''.
                

                TIME:       10:00 - 11:00 a.m.

         ROOM:          Davis Centre Room DC2305 (AI Lab)


         ABSTRACT

         In  the context of the vast and still rapidly expanding
         Internet, the problem of Internet information retrieval
         becomes  more  and  more  important.  Although the most
         popular at the moment, the keyword-based search engines
         are  just  one  piece in a complex software mosaic that
         needs  to  be  created  in  order  to  provide  a  more
         efficient and scalable solution.

         I try to show that the multi-agent approach is a viable
         methodology for this task, and how the natural language
         processing  could  be  used  in  it,  as well as why it
         should   be   used.   Two   implementations  and  their
         theoretical  foundations  are  presented:  One  is  the
         natural   language   parser  generator  NLP4InIR  which
         produces  parsers  in  C or Java; and, the other one is
         the  communication  part  of  the multi-agent framework
         MIN.  The  higher  levels  of  the  framework  are also
         discussed  and  a  demo implementation of a multi-agent
         system is presented.

10. Lab Cleanup (until 14:30 or 5 minutes)