CGL Meeting Minutes

Wednesday, June 14, 1995

Location:: DC 2303, the lab
Time:: 1:30 PM
Chair:: Wilkin Chau

0. Attendance

Richard Bartels, Leith Chan, Stewart Chao, Wilkin Chau, Eric Davies, Ed Dengler, Don Dragomatz, Nicholas Durand, Wolfgang Heidrich, Fabrice Jaubert, Rick Kazman, Nathan Konrad, Sandra Loop, Steve Mann, Michael McCool, Stephen Mereu, Andrew Park, Glenn Paulley, Randall Reid, Haroon Sheikh, Riston Tapp, Greg Veres, Julie Waterhouse, Ian Little

1. Adoption of the agenda - additions/deletions thereto.

2. Coffee Hour and Next Meeting:

Coffee hour this week:

Glenn Paulley

Coffee hour next week:

Fabrice Jaubert

Next Meeting

Time:: June 21, 1995
Location:: DC 2303, the lab
Chair:: Bill Cowan
Tech Presentation:: Leith Chan

Forthcoming: (list next 4 and trades)

Chairs:: Eric Davies; Ed Dengler; Don Dragomatz; Nicholas Durand
Tech presentations:: Stewart Chao; Wilkin Chau; Bill Cowan; Eric Davies

3. Technical Presentation:

Presenter:

Richard Bartels

Interval Spline Wavelets for Radiosity
======================================

We will review the basics of light interaction with a surface
point, the rendering equation, and the basic radiosity equation.
The expansion of radiosity in terms of basis functions and
the solution of the radiosity equation through the Galerkin
method will be covered.  Examples will be shown in which
a piecewise constant basis with spatial subdivision is
contrasted with the use of a higher-order basis.  The talk
will close with some current ideas about designing a
basis/wavelet pyramid that passes from piecewise constant
to higher-order and is specifically tuned for use in
radiosity computations.

4. General Discussion Items:

Stephen Mereu thanks everybody for helping him to conduct the experiment.
A CD-ROM version of the SIGGRAPH course notes will be ordered for this year. Richard will print out some of the important notes and put them into a binder.

5. Action List (remember to update AL_active)

U of T Visit

6. Directors Meeting:

1) The Fields Institute for Research will run a visualization software on Physeter in August.
2) What format should be ordered for the SIGGRAPH course notes ? ( see General Discussion Item )

7. Seminar(s):

Splines and Graphics Seminar Computer Graphics Lab
University of Waterloo

Date: Wednesday, June 14, 1995
Time: 4:30 PM
Place: William G. Davies Computer Research Centre, Room 2102

The Institute for Comput Presents a Colloquium on
Genetics-Based Learning and Generalization of Heuristics

by: Dr. Benjamin Wah
of: Department of Electrical & Computer Engineering, U. of Illinois Date: Wednesday, June 14, 1995
Time: 3:30 PM
Place: William G. Davies Computer Research Centre, Room 1302

Abstract:

Since credit assignment is difficult in these domains,  operators
for  composing  new heuristics are usually model-free, domain in-
dependent, and syntactic in nature.  The operators we  have  used
are genetics-based; examples of which include crossover and muta-
tion.  Learning is based on  a  generate-and-test  paradigm  that
maintains a pool of competing heuristics, tests them to a limited
extent, creates new heuristics from those that  perform  well  in
the past, and prunes poor ones from the pool.

Three important issues in learning heuristics  in  these  domains
are  (a)  performance  evaluation  of  heuristics,  (b)  rational
scheduling of experiments in testing  candidate  heuristics,  and
(c) generalization of heuristics to unlearned domains.  The first
and the third issues will be addressed in this talk.

Experimental results will be presented in learning
-new designs of blind filters,
-new heuristics in CRIS and GaTest for test-pattern generation in
VLSI testing,
-new parameters in TimerWolf, and
-new decomposition heuristics for branch-and-bound search.

Everyone is welcome. Refreshments served.