Seminar Archives 2002-2003

CS Seminar
Wednesday, July 9th
10:45pm in CS Room 221

Privacy in Statistical Databases

Luc Longpre

Nowadays, a lot of confidential data is collected in order to derive statistics necessary for research. In addition to proper access control mechanisms, we need to worry about how much confidentiality is lost when statistics are released. We review different approaches to release meaninful statistical information while protecting the confidentiality of individual records in the database.

CS Seminar
Friday, July 11
12:30-1:30, CS 321

Vector Coherence Spaces

Antonio Carlos da Rocha Costa

In this work we introduce the notion of vector coherence space as a combined domain-theoretic and algebraic structure designed to allow the modelling of physical computational processes, that is, computational processes that happen within a space-time framework. Vector coherence spaces are coherence spaces obtained over webs of vector-tokens, with some suitable coherence relation that guarantees the construction of well-defined vector-sets. The vector-space-like behavior of such structure comes from an additional algebraic structure that is introduced over the order-theoretic structure. The final result is that vector coherence spaces and their transformations constitute a semantical category, where we hope to give denotations for computational processes which are tied to physical spaces.

CS Seminar
Tuesday, July 8
10:40-11:50 in CS Room 221

Simplicial Domains to Model Uncertainty in Measuring Processes

Gracaliz P. Dimuro

This work provides a general domain-theoretic model for the uncertainty underlying measuring processes, with a geometric interpretation based on simplicial complexes. This was done by establishing an analogy between the notions of approximation and computation in domain theory and the notion of measuring process performed with a measuring instrument. In this sense, a qualitative domain of coherent simplexes of instrument readings is introduced. Following the same approach, a qualitative domain of coherent simplicial complexes, adjoint to the former domain, is defined to model the histories of the measuring processes. We give an application of the theory to the modelling of simple perception modules in autonomous robots, and hint on the modelling of perception modules with sensor data fusion.

CS Seminar
Wednesday, July 2nd
10:45pm in CS Room 221

A New Algorithm for Computing Page Rank for Google Searches

Vladik Kreinovich

PhD Dissertation Defense
Wednesday, May 7th
4pm in CS Room 221

Building Knowledge Systems in A-Prolog

Monica Nogueira

This work is written in the context of the logic-based approach to Artificial Intelligence (AI) proposed by John McCarthy in 1959. According to this approach an agent should have knowledge of its world and its goals, and the ability to use this knowledge to infer its course of action. This logic-based method suggests that a mathematical model of an agent should contain a formal language capable of expressing commonsense knowledge about the world, a precise characterization of valid conclusions which can be derived from theories stated in this language, and a means which will allow the agent to arrive at these conclusions.

The purpose of this dissertation is to investigate the applicability of one such language, A-Prolog, for the development of medium size knowledge-intensive systems. A-Prolog is a declarative logic programming language based on stable models/answer sets semantics of logic programs. It allows the representation of defaults and multiple interesting aspects of reasoning about actions and their effects. There is a recently developed methodology of representing knowledge in A-Prolog. There are also rather efficient inference engines associated with the language. Our goal was to test this methodology and these inference engines on sizeable engineering applications.

In this dissertation, we developed two such applications. The first is a small system, designed as a classroom tool for teaching digital circuits which allows the functional and behavioral representation of these circuits at the gate-level of abstraction. The second is a substantially larger application - the implementation of a decision support system for the Space Shuttle flight controllers. This work involved a representation of a substantial amount of knowledge about the shuttle as well as the execution of complex planning (and other reasoning) tasks. The project was successful and the system is now in the hands of USA - the company responsible for overseeing the operation of the Space Shuttle.

The thesis describes design and implementation of these systems and discusses some lessons derived from this experience. We believe that the lessons can be of interest to AI researchers working in the areas of knowledge representation, non-monotonic reasoning, and planning, as well as to software engineers involved in the construction of knowledge intensive systems.

Colloquium
Friday, April 11
10:45--11:45 in CS Room 221

An Open Framework for Developing Distributed Computing Environments for Multidisciplinary Computational Simulations

Purushotham Bangalore, Mississippi State University

Multidisciplinary computational simulations involve interactions between distributed applications, datasets, products, resources, and users. Because the very nature of the simulation software emphasizes a single-computer, small-usership and audience, the kinds of applications that have been developed often are unfriendly to incorporation into a distributed model. However, advances in networking infrastructure, and the natural tendency for information to be geographically distributed place strong requirements on integration of single-computer codes with distributed information sources, as well as multiple computer codes that are geographically distributed in their execution. Thus, it is evident that there is a need for distributed computing environments suitable for users of multidisciplinary simulations to connect disparate applications, datasets, products, resources, and users into a single user-friendly system while hiding the complexities of distributed computing from the users of this system.

The Enterprise Computational Services (ECS) framework, based on the Java 2 Enterprise Edition (J2EE) multi-tier architecture, Web Services, and Grid Technologies, provides a reusable and extensible infrastructure for connecting standalone applications into a single user-friendly system while reducing complexities of multi-user, multi-site, multi-program, multi-resource, and multi-organizational computing. Several distributed computing environments are developed using the ECS framework at the Engineering Research Center at Mississippi State University. One such environment, the Distributed Marine Environment Forecast System (DMEFS), is described to illustrate how the ECS framework could be used to develop distributed environments that integrate disparate users, resources, and applications while improving the productivity of users.

Students-Only Seminar
Friday, April 11
14:45-15:45 in CS Room 221

Introduction to Grid Computing

Purushotham Bangalore, Mississippi State University

Over the last few years Grid Computing has gained popularity as the emerging architecture for next-generation high performance computing. Grid Computing aims to provide ubiquitous access to distributed HPC resources shared between multiple organizations and provide "virtualization" of computational resources. In this talk, a brief introduction to Grid Computing is provided. The benefits of Grid Computing along with the issues and challenges that must be addressed are also presented. Potential impact of Grid Computing on HPC applications and application development is also discussed. Differences and similarities between Cluster Computing and Grid Computing are also explained.

Colloquium
Friday, April 11
1:30-2:30 in CS Room 221

Bodily Expression and `Informatical' Expression:
Analysis and Synthesis of Human-Robot Non-verbal Interaction

Toru Nakata, National Institute for Advanced Science and Technology

Non-verbal expressions are important in our daily communications, yet they are difficult to handle in computers. This talk will describe my efforts to realize computationally usable expressions of them. My first challenge was to create a quantitative analysis method on body movement expressions in humans and animal-shaped robots. I adopted the choreographic techniques of Rudolf Laban, who had considered the relationships between expressional meanings and dynamical features of human body motions. This theory was used in software for estimating users' impressions of a robot's motion from it motion paths. However, any bodily expression cannot avoid being boring when it is repeated many times. This implies that effective bodily communication in interaction is also dependent on the newness of the expressional signals. I therefore developed mathematical tools to measure and control the newness of interactive expression.

Dr. Nakata earned his doctoral degree in 2001 in Mechano-Informatics at the University of Tokyo. He is currently a researcher at the Digital Human Research Center, a new branch of the National Institute for Advanced Science and Technology (AIST), which is affiliated with the former Electro-Technical Laboratory.

Colloquium
Friday April 4, 2003
11:15 in Room CS 221

Processor Allocation in Full-Cube Multicomputers

Students-Only Seminar
Friday April 4, 2003
3:15 in Room CS 221

Parallel Computing with Intel Microprocessors

Abstract: We'll look at recent offerings in the IA-32 and IA-64 microprocessor product lines. Hyperthreading (simultaneous multithreading), trace cache, EPIC/VLIW, control and data speculation, predication, parallel compares, and multi-way branches are some of the adopted mechanisms to facilitate parallel computing and achieve higher performance. The talk will touch on future supercomputers and the design feasibility of 1 billion transistor processors.

TRACS Seminar, UTEP Computer Science
Friday April 4, 2003
12:30-1:30 in Room CS321

Professor Weldon A. Lodwick, of the Department of Mathematics at the University of Colorado at Denver, will give a talk titled:

Being Certain About Uncertainty

Abstract: An approach to obtain bounds on functions of variables and parameters whose values are uncertain will be presented. The theory is developed and examples will be given. These methods lead to methods for arithmetic on distributions and applications to situations where Monte Carlo methods are typically applied. How to use these methods in optimization under uncertainty and risk analysis will be shown. However, the focus of this presentation will be on the theory, algorithm, guaranteed bounds, approximation from the bounds and examples.

coordinator: Dr. Vladik Kreinovich

February 10, 10:30, CS 221

Dr. Yuri Gurevich, leader of the Foundations of Software Engineering group at Microsoft Research, will give a talk titled:

Executable Specifications: The ASM approach

Some people think that executable specification is a contradiction in terms. We think that executable specifications will change the way software is designed, developed, tested and documented. Our opinion is based on the theory of abstract state machines, extensive international experimentation with ASMs, and the applied ASM work of our own group, the group on Foundations of Software Engineering.

Contrary to natural sciences, computer science is about artificial reality, about computer systems. Mathematically speaking, what is a computer system? A computer system may have many meaningful levels of abstraction. Fix such a level. The ASM theory tells us that there is an abstract state machine that, behaviorally, is identical to our system on the chosen abstraction level. The specification language AsmL, developed by the FSE group, makes writing ASM models practical. Our tools allow the developers (more and more) to experiment with their design, validate it and enforce it. The tools allow testers to be involved earlier in the software development cycle and empower them to test the intended functionality of software (and not only its robustness).

Bio: Yuri Gurevich is Senior Researcher at Microsoft Research in Redmond, WA. He is also Professor Emeritus at the University of Michigan, ACM Fellow, Guggenheim Fellow, and Dr. Honoris Causa of Limburg University in Belgium. In Microsoft, he heads the group on Foundations of Software Engineering.

coordinator: Dr. Francois Modave, Tel: (1 915) 747-5564

Computer Science Seminar, UTEP
Monday Feb 17, 2003
11.00 in Room CS221

Yoonsik Cheon, PhD candidate at Iowa State University, will give a talk titled:

A Runtime Assertion Checker for the Java Modeling Language

Debugging is made difficult by the need to precisely describe what each piece of the software is supposed to do, and to write code to defend modules against the errors of other modules; if this is not done it is difficult to assign blame to a small part of the program when things go wrong. Similarly, unit testing also needs precise descriptions of behavior, and is made difficult by the need to write test oracles. However, debugging and testing consume a significant fraction of the cost of software development and maintenance efforts. Inadequate debugging and testing also contribute to quality problems.

In this talk, I describe a runtime assertion checker for the Java Modeling Language (JML) that helps in assigning blame during debugging and in automatic generation of test oracles. JML is a formal behavioral interface specification language for Java. And its runtime assertion checker represents a significant advance over the current state of the art, because it can deal with very abstract specifications which hide representation details, and other features such as quantifiers, interface specifications, and inheritance of specifications. Yet JML specifications have a syntax that is easily understood by programmers. Thus, the runtime assertion checker has the potential for decreasing the cost of debugging and testing.

This talk is based on a joint work with Dr. Gary T. Leavens at Iowa State University. The work is supported in part by the NSF under grant CCR-0097907 and CCR-0113181.

Biographical Sketch: Yoonsik Cheon is currently a Ph.D. candidate in the Department of Computer Science at Iowa State University. Since 1995, he has been working for Electronics and Telecommunications Research Institute (ETRI) of Korea as a member of technical staff. He holds a MS in Computer Science from Iowa State, and a BS in Computer Science from Korea University. His primary research interest is in formal methods, in particular, practical application of formal specifications into object-oriented programming.

coordinator: Dr. Francois Modave, Tel: (1 915) 747-5564

Computer Science Seminar, UTEP
Wednesday February 5th, 2003
10.30-11.30am in Room CS221

Dr. Juris Reinfelds, Professor of ECE, NMSU, Emeritus will give a talk titled:

The Past, Present and Future of a Programmer's Theory of Programming

Edsger W. Dijkstra (1930-2002), contributed a wealth of concepts, methods and insights to the science of programming. To commemorate the passing of a friend and colleague, the first part of my colloquium will contain a brief retrospective of Dijkstra's work, especially his theory of programming that we can call a "programmer's theory" in contrast to the "mathematician's theories of computer science. To show that a valuable past is a great prologue, the second part of the colloquium will discuss a new programmer's theory of programming, currently under development as the Kernel Language approach to programming, which brings to light many historical accidents that can be discarded and shows how we can simplify distributed programming by looking at programming in a new light.

coordinator: Dr. Francois Modave, Tel: (1 915) 747-5564

Computer Science Seminar, UTEP
Wednesday November 20, 2002
10.30-11.30am in Room CS221

Dr. Brian d'Auriol

Secure Hardware Level Architectures

Computer security includes overlapping controls in an attempt to provide information privacy, integrity and availability. Many of the common controls include authentication, encryption, firewalls and physical lock-downs. Despite these efforts, security issues at the computer architecture and hardware levels are often overlooked in today's heightened visibility of network related attacks. Never-the-less, secure architectures have been proposed for many decades. In 1994, NIST established the FIPS PUB 140-1 for the security requirements for cryptographic modules; the definition of which covers the architecture hardware components of such a module. Most recently, some secure architectures have been proposed with claims of meeting all of the Level 4 requirements (the highest level of security) of the FIPS 140-1 standard. This talk presents a number of information privacy, integrity and/or availability issues at the architectural level. Some historical comments about secure architectures, especially, the role of encryption modules are discussed. Recent work in the field is described. The FIPS 140-1 and other relevant standards are briefly surveyed. If time permits, comments relating secure architectures to the National Strategy to Secure Cyberspace For Comment Draft (a Sept. 2002 document by the President's Critical Infrastructure Protection Board) will be given.

Computer Science Seminar, UTEP
Wednesday November 6, 2002
10.30-11.30am in Room CS221

Dr. Luc Longpré

Passwords and Access Control

Passwords are a way of identifying the user of a computer in order to control its access. We review ways in which computers handle this task, the common attacks on trying to get around this control and ways to handle passwords on several machines. We also give an overview of methods other than passwords: smartcards and biometrics.