SaS Seminars 2003
Software and Systems Research Seminar Series
Functional Reactivity: Eschewing the Imperative
Date: December 12, Place: Alan Turing (Estraden) Time: 10.15
Henrik Nilsson, The University of Nottingham, School of Computer Science an Information Technology
Abstract Functional Reactive Programming (FRP) is a paradigm for reactive programming in a pure functional setting. FRP originated in Conal Elliott's and Paul Hudak's work on Functional Reactive Animation (Fran), and has since evolved in a number of different directions and into a number of different concrete implementations. FRP share many features with, on the one hand, synchronous languages, like Esterel and Lucid Synchrone, and, on the other, languages for modelling and simulation, like Simulink and Modelica. However, FRP has a number of distinct features that set it apart, for example its support for describing reactive systems with highly dynamic structure, and the fact that reactive components are first class entities, yielding considerable meta-programming capabilities for free. This talk gives an overview of Functional Reactive Programming in the context of Yampa, the latest FRP implementation from Yale University.
Declarative Multi-Paradigm Programming in Curry
Date: November 27, Place: Alan Turing (Estraden) Time: 15.15
Michael Hanus, Institut für Informatik und Praktische Mathematik, Christian-Albrechts-Universität zu Kiel
The multi-paradigm language Curry is an attempt to combine in a seamless way features from functional programming (reduction of nested expressions, higher-order functions, lazy evaluation, polymorphic type systems), logic programming (logical variables, partial data structures, constraint solving, built-in search), and concurrent constraint programming (concurrent evaluation of constraints with synchronization on logical variables). Thanks to this combination, Curry provides new programming techniques (e.g., demand-driven search) and better structuring facilities, and it avoids impure features of logic languages like Prolog. The development of Curry is an international initiative intended to provide a common platform for the research, teaching and application of declarative multi-paradigm languages.
This talk provides an introduction to Curry and related multi-paradigm programming techniques. It will be shown that functional languages can be easily extended to cover features for logic and concurrent programming. Furthermore, we discuss some advanced features of Curry for constraint or distributed programming. Finally, we demonstrate the advantages of such a multi-paradigm language by some applications, e.g., the high-level implementation of web-based systems.
"Weaving the Adpative Web": Towards personalization in the Semantic Web
Date: October 23, Place: Alan Turing (Estraden) Time: 15.15
Universität Hannover, Institut fuer Informationssysteme,
Member of the L3S: Learning Lab Lower Saxony
Chair of ABIS, the Working Group "Adaptivity and User Modelling in Interactive Systems" of the Geraman Society for Computer Science (GI).
The idea of the Semantic Web to give information a "well-defined meaning, better enabling computers and people to work in cooperation" rises the question of an Adaptive Web which knows like a personal agent the specific requirements of a user, takes goals, preferences, or the actual context into account in order to optimize the access to electronic information.
Desing and formation of an Adaptive Web are just in the beginning. Fundamental research for personalized access, retrieval and presentation of information can be found in the area of recommender systems (and their principal application area "e-commerce"), and in the area of adaptive hypermedia (and their principal application area "e-learning"). In this talk, we investigate the applicability of methods and techniques from adaptive hypermedia for designing an "Adaptive Web". A logic-based definition of adaptive hypermedia will be presented, which supports encapsulation and re-usability of adaptation techniques. This characterization allows for formulation and embodiment of personalization techniques on the logic-level of the Semantic Web Tower. A draft design for an Adaptive Web, based on rule-based reasoning enabled by Semantic Web technologies, will be presented.
Traces of Randomized Distributed Algorithms As Markov Fields
Date: September 25, Place: Alan Turing (Estraden) Time: 15.15
Directeur de recherche, CNRS, at Ecole Normal Superieur de Cachan
Assistant Director of LSV (the Laboratory for Specification and Verification)
(in cooperation with S. Messika and C. Picaronny)
Given a randomized distributed algorithm A that operates on a ring (or linear array) of N processes, we use the fact that the traces of computation via A form a Markov field (i.e., a random field governed by local interactions), and that A has a unique stationary distribution if the Markov field has no phase transition. We then use van den Berg's sufficient condition on ``paths of disagreement'' for guaranteeing the absence of phase transition of Markov fields, as an original tool for proving the uniqueness of the stationary distribution D and the rapid mixing of A towards D. This new method of proving rapid mixing is illustrated on Herman's self-stabilizing algorithm of mutual exclusion and the classical random walk on a ring.
The Next-Generation Web: Information in the Third Millenium
Date: June 5, Place: John von Neumann (Belöningen), Time: 15.15
W3C, Research Scientist at the MIT,
Professor, Department of Computer Science, University of Venice
The World Wide Web has grown, and it size is now enormous. But if its size and the number of people using it is impressive, it is also true that its original architecture is showing all its limitations. Information handling has become more and more difficult, and the enormous potential of the Web seems yet unexpressed. In this presentation we will show what are the solutions that are under development and standardization to bring the Web further into the future, so to lead the Web to its next generation.
Introduction to the Protein Secondary Structure Prediction (PSSP): a Model and a Metaclassifier
Date: May 15, Place: Estraden, Time: 15.15
Anita Wasilewska: Professor, Department of Computer Science, State University of New York at Stony Brook
The talk will discuss a joint work with Dr. Victor Robles Forcada of the Technical University of Madrid. We provide an intuitive introduction to the protein secondary structure prediction research preliminaries and problems. We explain what are the proteins in real life, provide a simple mathematical model, give an overview of some of the newest supervised classification methods and classifiers and discuss their results. The classifiers, of the third generation, are chosen from 10 available servers on the Internet. Finally, we present ideas behind our Metaclassifier based on Bayesian Network approach. The Metaclassifier is currently reaching an excellent accuracy prediction level (> 81%).
Compilation for Embedded Processors
Date: April 29, Place: Estraden, Time: 15.15
Y.N. Srikant, Professor and Chairman, Dept. of Computer Science and Automation, Indian Institute of Science
With the advent of system level integration and system-on-chip, the centre of gravity of the computer industry is now moving from personal computing into embedded computing. The resulting upheaval is only just beginning to be widely appreciated. Embedded software is pervasive and appears in vehicles, telephones, audio-video-equipment, aircraft, toys, security systems, medical diagnostics, weapons, pacemakers, climate control systems, manufacturing systems, intelligent power systems etc.
Every embedded system very often contains a DSP (digital signal processing) processor. Most embedded systems have constraints on memory size, power, and overall dimensions. However, their computing requirements are by no means low. Code running on such processors must necessarily be fast, compact and also energy-efficient. Compilers for embedded processors are difficult to design and implement. They must generate code which utilizes several features of embedded processors such as, clusters of multiple functional units, dual memory banks, large number of registers, low power operation, special instructions, etc., very efficiently. These pose interesting challenges to a compiler designer.
In this talk, we survey the techniques available for efficient code generation for embedded processors and outline the ongoing research in this area.
Professor Y.N. Srikant received his Ph.D in Computer Science from the Indian Institute of Science in 1986. He is currently a Professor Chairman of the Department of Computer Science and Automation at the Indian Institute of Science. His areas of interest are compiler design and tools for software design. He is one of the editors of a handbook of compiler design published by CRC Press in September 2002. Dr. Srikant is the recipient of the Young Scientist medal of the Indian National Science Academy. More details regarding his publications may be reached here.
System development with Action systems
Date: March 13, Place: Estraden, Time: 15.15
We present an approach to system development using the Action Systems formalism as the underlying formal framework. The approach is based on viewing a system, consisting of hardware and/or software, as a set of concurrent, distributed, co-operating entities. We exemplify the development approach via a few case studies on the design of e.g. control systems and software and mobile agents.
Hierarchical Approaches to Digital Test
Date: February 12, Place: Estraden, Time: 15.15
Raimund Ubar, Professor, Technical University of Tallinn
An overview of research fields at the Test Lab in Computer Engineering Dept. of TU Tallinn will be presented. Then a popular introduction to the main problems in the field of Digital Test will follow. A special attention will be concentrated on the hot question: how to improve the testing quality at increasing complexities of today's systems? As a promising solution, the approach of combining high-level modelling with defect orientation will be discussed. Next, a short introduction to Decision Diagrams (DD) will be given. DDs serve as a good tool for hierarchical modelling of defects in digital systems. It will be shown how DDs help to achieve efficient procedures of hierarchical test generation and fault simulation in digital systems. Finally, an overview of testing tools developed at TU Tallinn together with corresponding experimental results will be presented.
Raimund Ubar is a professor at TU Tallinn, and one of the three Estonian research professors elected recently for three years by Estonian Academy of Sciences. His research interests are fault modeling, simulation and diagnosis, test generation, design for testability and fault tolerance in digital systems. He has lectured and given talks at universities of more than 10 countries, participated during the last 10 years in 8 European projects, published more than 150 papers and contributed yearly for Program Committees of about 10 international conferences. He is a member of Estonian Academy of Sciences.
AN EXAM EXAMINED
A discussion on knowledge categorization applied to the Occasional Programmer's sphere.
Date: January 30, Place: Estraden, Time: 15.15
Olle Willen, IDA
Occasional Programmers are believed to demonstrate a more pragmatic approach to the knowledge area of programming than professionals or academics. Therefore the classes of qualitative knowledge they account for may prove to be based on their pragmatic intentions. Classical taxonomies of cognitive abilities emanate from the constructivists' theories that new abilities are built upon the already existing, stepwise leading to higher cognitive qualities. This is perhaps contradictory to the way the Occasional Programmer is learning, since it is assumed that he strives to keep away from too hard cognitive challenges. This paper is based on observations of written examinations from a group of Occasional Programmers. On the hypothesis that they perhaps reveal a different style of learning an experimental model of knowledge categories is designed and evaluated. Attempts are then made to categorize the same examination in accordance with established taxonomies (Bloom, SOLO) and the results studied. Our intention is to carry out experiments with these taxonomies, and in connection with the results discuss the concept of knowledge categorization in order to get better acquainted with the problem area - an attempt to approach the secrets of knowledge consctruction. It is concluded that many problems are connected to the use of models of knowledge categorization, but also that a lot could be learned from such use that will enhance a teacher's awareness of learning.
Page responsible: Christoph Kessler
Last updated: 2012-08-17