Title: Reflectance mapping and fundamental componentEach reflectance can be separated into two components - a fundamental component and metameric black component. The fundamental component is unique and identical for all metameric reflectances (reflectances that produce the identical colour sensation). The metameric black component does not affect any colour sensation. It has tristimulus values of ( 0, 0, 0 ). In this talk, the study of reflectance mapping using fundamental component will be discussed. A brief comparison between this mapping and the projection mapping will be shown.
Splines and Graphics Seminar
Is there one?
SYMBOLIC COMPUTATION SEMINAR
Who: E.V. Zima, Dept. of Comp. Sci., Moscow University
Title: Symbolic-numeric cocktail
Date: Wednesday, July 5, 1995
Time: 3:30-4:30 p.m.
Room: DC 1304
Abstract:
Different problems of symbolic-numeric interface
implementations will be considered (with much attention
paid to "single user-single process" MS DOS like
architecture):
- implementation and use of static and dynamic models
of symbolic-numeric interface for MS DOS computers;
- syntactic-oriented approach to design, analysis and
modification of numeric and symbolic programs;
- use of Compter Algebra systems as the tool for
source-to-source numeric program optimization.
All parts of the talk are supported by corresponding
demonstrationss (the AlgInt package for Turbo-Pascal,
structural Maple and Turbo-Pascal editors, etc.)
ICR DISTINGUISHED VISITOR SEMINAR
Who: Dr. Kiyoo Itoh, Central Research Laboratory,
Hitachi Ltd., Tokyo, Japan
Title: VLSI Memory Design
Date: Thursday, July 6, 1995
Time: 2:30 pm.
Room: Davis Centre, Room 1304
Abstract:
VLSI memory design is reviewed in terms of the key circuit-design
issues with emphasis on DRAMs. This talk will include the fol-
lowing issues: (i) High Signal-to-Noise Ratio (S/N) Circuits:
cell design and S/N determinants, cell signal charge increase,
comparison between trench capacitor and stack capacitor, and
capacitor electric field. (ii) High Speed Circuits: current flow
deconcentration and small signal transmission. (iii) Low Voltage
Circuits: crisis for CMOS LSIs due to sub threshold-current in-
crease, and switched-source impedance scheme and its applica-
tions. Based on the presentation, a perspective will be given
with clarifying challenges to ultra-low voltage RAM design.
SYMBOLIC COMPUTATION SEMINAR
Who: Alain Sausse, INRIA-Sophia Antipolis, SAFIR project
Title: Distributed Software Architecture. Application to compute
the Primary Decomposition of Ideals
Date: Thursday, July 6, 1995
Time: 4:00-5:00 p.m.
Room: DC 1304
Abstract:
There is a large family of systems for symbolic
computation. Most of them have libraries of algorithms
for the resolution of complex mathematical problems.
From general purpose systems to specialized packages,
the user can find the suitable tools to solve his
problems. Nevertheless, in many cases the user wants to
deal with more than one of these tools at once to get
the best of them all. There is no simple way to make
systems collaborate in this way except using files and
ad-hoc tools to convert data from one system to
another.
Thus, we designed a software component the Central
Control that enables several symbolic systems to work
together by exchanging data. The Central Control has
been designed to be the kernel of an environment for
scientific computations which provides a common access
to various tools developped by the PoSSo project
together with access to other systems like, for
example, Maple and Macaulay. The Central Control
----- --------
provides basic mechanisms to ensure term translations
in order for servers to interoperate and exchange data
despite the difference of syntaxes and semantics. The
Central Control achieves its goals by requiring as
little as possible from the tools and in its first
version by using a particular programming language to
provide a unified view for the objects and the
operations performed by the connected tools. In its
current version, it is a Scheme interpreter, extended
------
with special primitive operations and types. We
describe the main features of the Central Control
(terms, bindings, servers, promises, handles,
rewriting) and show how the Central Control can be an
effective tool to study a problem coming from celestial
mechanics and then, to compute primary decompositions
of ideals.
SEMINAR
Who: Dr. Dawn Jutla, Dept. of Comp. Sci.,
Technical University of Nova Scotia
Title: The Multiview Memory Model
Date: Friday, July 7, 1995
Time: 10:30 am.
Room: DC 1304
Abstract:
This talk first explores the issues surrounding the motivation
and development of the Multiview On-the-fly Memory Model (MOM)
memory model. Currently many programmers effectively bypass the
regular operating system services by implementing more suitable
resource management services in their user level applications.
The MOM memory model and its supporting architecture provides for
the customization of OS access control services; this avoids the
duplication of services across applications. Customizability of
applications is provided through MOM's support of multiple access
control protocols per region of memory.
A significant feature is the model's provision of variable sized
units for the purposes of access control (e.g. locking), where
the control units can be of a size different from the system page
size. How variable sized protection units can be achieved
without affecting the system's page size, thereby bypassing many
of the problems associated with OS system support for architec-
turally provided variable sized pages (e.g in MIPS R4000), is
shown.
Reduction in context switching may be achieved in two ways:
through the implicit synchronization of access to memory loca-
tions and by dedicated cache support for access control proto-
cols. Even though the cache support is not absolutely necessary
to support the memory model, the reduction in data cache pollu-
tion and the increased effectiveness of the TLB are arguments for
the expense of dedicated cache support. This research represents
work-in-progress towards the fulfillment of the PhD requirements
at the Technical University of Nova Scotia, Halifax.
SOFTWARE ENGINEERING SEMINAR
Who: Jan Madey, Warsaw University, Poland
Title: Trace Assertion Method and Finite State Machines
Date: Friday, July 7, 1995
Time: 1:30-2:30 p.m.
Room: DC 1304
Abstract:
The trace assertion method (in short: TAM) is a formal
method for abstract specification of interfaces of
software modules being designed according to the
"information hiding" principle. A module implements a
number of homogeneous, independent objects. A trace
specification is a black-box specification, i.e., it
describes only those features of an object that are
externally observable.
The method was introduced by W. Bartusek and D.L.
Parnas some 15 years ago and since then has undergone
many modifications. In recent years there has been an
increased interest in TAM. Software tools supporting
practical usage of TAM for software engineering
projects are under development, the method is being
tested on different applications, its foundations are
being studied.
In this talk we will discuss some fundamental concepts
of TAM and present a model for trace specifications
(Mealy machine) together with an argument about the
expressiveness of TAM. In particular, we will
explicitly distinguish two equivalence relations on the
set of traces: one, based on the observable features of
an object, and one, defined by a trace specification of
this object.
MASTER'S ESSAY PRESENTATION
Who: Jilong Chen, graduate student, Dept. of Comp. Sci.,
University of Waterloo
Title: Gesture Recognition and Application
Date: Thursday, July 13, 1995
Time: 3:30-4:30 p.m.
Room: DC 1331
Abstract:
As human beings we use gestures everyday. Many human-
computer interactions can be more naturally expressed
as gestures.
Currently we have two major user interfaces: command
language based user interface and graphical user
interface. They do good job in most conventional
applications. However, it is hard to use them in 3D
applications. Gesture based user interfaces use hand
and body gestures as commands ,they are more powerful
to express user's intentions.
Gesture recognition can be thought of as a special
pattern recognition task. There are two elements
involved: one is how to measure each pattern, the other
is how to use these numbers to recognize them. We will
review the work done by researchers, point out their
pros and cons.
We will propose our approach in both measuring and
recognizing: we applied Fourier transformation as well
as wavelets transformation to the raw data, and applied
established pattern recognition method. The test
results showed ours has a higher recognition rate than
those published. We will also propose a pipeline for
building gestural user interface. We built a model in
X-Window environment, which collects raw data by
sampling the positions of mouse cursor. Finally, we
are going to discuss industry products that apply
gesture recognition techniques, possible applications,
architectural support for gestural user interfaces,
information representations, interaction techniques,
etc.