Meeting Agenda

Wednesday, June 25^th, 1997

Location:

DC 1304

Time:

13:30

Chair:

Richard Bartels

1. Adoption of the Agenda - additions or deletions

2. Coffee Hour

Coffee hour this week:

???

Coffee hour next week:

???

3. Next meeting

Date:

July 2^nd, 1997

Location:

DC 1304

Time:

13:30

Chair:

Wilkin Chau

Technical presentation:

Richard Bartels

4. Forthcoming

Chairs:

1. Blair Conrad (7/9)
2. Bill Cowan (7/16)
3. Itai Danan (7/23)

Tech Presenters:

1. Wilkin Chau(7/9)
2. Blair Conrad (7/16)
3. Bill Cowan (7/16)

5. Technical Presentation

Presenter:

Balasingham Balakumaran

Title:

TBA

Abstract:

TBA

6. General Discussion Items

7. Action List

• Japanese Visitors
• Alias Visit Visit

8. Director's Meeting

9. Seminars

• The Complete Works of John von Neumann (Abridged)
  
  Allan Donsig
  Wednesday, June 25, 1997
  3:30 pm
  MC 4041
  ABSTRACT
  John von Neumann (1903-1957) is, arguably, the greatest mathematician of the twentieth century. He made major contributions to most of the subjects represented in the math faculty, including computer science, statistics, and many parts of mathematics, pure and applied. I will sketch his life, explain a few of his theorems and, I hope, give an idea of what he was like. For the theorems, some linear algebra would be helpful.



• CSGSA Celebrity Lecture Series
  Computational Genetics, Physiology, Metabolism,
  Neural Systems, Learning, Vision, and Behavior
  or
  PolyWorld: Life in a New Context
  
  Larry Yaeger
  Wednesday, July 2nd at 2:30 in DC1302
  ABSTRACT
  I will discuss a computer model of living organisms and the ecology they exist in, called PolyWorld. PolyWorld attempts to bring all the principle components of real living systems into a single artificial (man-made) living system. PolyWorld brings together biologically motivated genetics, simple simulated physiologies and metabolisms, Hebbian learning in arbitrary neural network architectures, a visual perceptive mechanism, and a suite of primitive behaviors in artificial organisms grounded in an ecology just complex enough to foster speciation and inter-species competition. Predation, mimicry, sexual reproduction, and even communication are all supported in a straightforward fashion. The resulting survival strategies, both individual and group, are purely emergent, as are the functionalities embodied in their neural network "brains". Complex behaviors resulting from the simulated neural activity are unpredictable, and change as natural selection acts over multiple generations.

  In many ways, PolyWorld may be thought of as a sort of electronic primordial soup experiment, in the vein of Urey and Miller's classic experiment, only commencing at a much higher level of organization. I also see PolyWorld as a means to begin exploring Artificial Life as a path toward Artificial Intelligence, utilizing the same key elements that led to natural intelligence: the evolution of nervous systems in an ecology.

  I will discuss the design principles employed in PolyWorld, the "species" that have evolved in various simulations, and the group and individual behaviors observed in these simulations. I will also at least briefly address the difficult issues around deciding when "Life" is actually present in a system.




• DEPARTMENT OF COMPUTER SCIENCE
  UNIVERSITY OF WATERLOO
  SEMINAR ACTIVITIES
  HARDWARE VERIFICATION SEMINAR
  Thursday, June 26, 1997
  
  Radu Negulescu
  Formal Verification of Digital Circuits by FIREMAP.
  TIME: 10:30-11:30 a.m.
  ROOM: DC 3301
  ABSTRACT
  It is by now notorious that a design flaw in a digital circuit can entail huge penalties (the famous Pentium bug is reported to have cost US$ 500 million).

  We present FIREMAP, a tool for analyzing concurrent systems, and we describe some applications of this tool to the formal verification of digital circuits. The tool and formalism have a high degree of flexibility, as the underlying theory has been built with an abstract notion of an execution, and without any restrictions on the concurrent processes that participate in compositions or comparisons. In circuit applications, we have been able to deal with synchronous and asynchronous circuits, with general CMOS circuits, with terminals that are sometimes inputs and other times outputs, with transistor-level circuits where the transistors are approximated by switches, and with delay constraints and isochrony. We have tested FIREMAP on several small industrial designs. Neither the tool nor the underlying theory are limited to circuits.

  A common problem in formal verification is that the number of states may grow exponentially with the number of components ('state explosion'). To deal with large state spaces, FIREMAP combines a BDD-based implementation (using a library provided by the Model Checking group at Carnegie Mellon University) with full support for modular and hierarchical verification, and, in some cases, with support for partial order techniques that may reduce the number of states that need to be verified.




• DEPARTMENT OF COMPUTER SCIENCE
  UNIVERSITY OF WATERLOO
  SEMINAR ACTIVITIES
  SCIENTIFIC COMPUTATION SEMINAR
  Thursday, June 26, 1997
  
  P.L. George , INRIA, Rocquencourt, Paris
  Delaunay triangulation and meshing.
  Applications to isotropic and anisotropic adaptive mesh construction
  TIME: 3:30-4:30 p.m.
  ROOM: DC 1304
  ABSTRACT
  Delaunay triangulation algorithms are of great importance for (finite element) meshing purpose. Adaptive meshing can be based on modified Delaunay algorithms both for isotropic meshing where element size are prespecified and for anisotropic meshing when sizes as well as directional features are demanded. In this talk, we will discuss how to extend the a priori well known Delaunay method so as to achieve adapted meshes suitable for finite element computations.



• DEPARTMENT OF COMPUTER SCIENCE
  UNIVERSITY OF WATERLOO
  SEMINAR ACTIVITIES
  MASTER'S THESIS PRESENTATION
  Thursday, June 26, 1997
  
  Najmeh Joze-Khajavi
  Analysis of Techniques for Clustering Large Data Sets
  TIME: 10:00-11:00 a.m.
  ROOM: DC 1331
  ABSTRACT
  This research involves investigation of clustering techniques for large data sets. Clustering of large data sets can be counted as one of the problems in data mining. We focus on methods of clustering that scan the data set only one time. This is due to the fact that scanning large data sets is a long process. In this thesis, a two phased method combining random sampling and an iterative clustering method LBG (Linde, Buzo, Gray) called CLURS (CLUstering based on Random Sampling) is proposed. CLURS provides an abstract representation of the large data in main memory and LBG uses this representation and provides the final clusters. It is shown that this method outperforms an existing technique named BIRCH (Balance Iterative Reducing and Clustering using Hierarchies). BIRCH also has two phases. However, in its first phase it uses a tree structure to represent different clusters. The advantages of our method are its efficient use of memory and its fast execution time.

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