International Multiconference
on Computer Science and Information Technology

18-20 October 2010, Wisła, Poland
logo PTI

Polish Information Processing Society
PRINT article  SEND link to article  facebook google delicious dig 

Keynote Speakers

gentzsch_wolfgang.jpgWolfgang Gentzsch

Over the last 50 years, computational sciences and engineering have advanced the development of new and ever faster and better computers, algorithms, and software tools, and vice versa. But the better the technologies we build the more demanding our scientific and engineering applications. That's why optimization is everywhere, et the level of technologies and computer architectures, at the middleware level, at the algorithm and the engineering application level, and even at the level of computing paradigms, where we came from mainframes, to vector and parallel computers, to grids and, recently, to clouds. We successfully optimized our algorithms and mapped them to the underlying architecture, e.g. with overlapping communication with computation, better load balancing through domain decomposition into parallel processes, or using library  routines optimized for the specific architecture and processors.

Still many scientific and engineering applications are extremely performance-hungry, or they simply waste a lot of computing cycles and memory. The two alternatives to cope with these challenges are: always using the largest and fastest (and most expensive) supercomputers for jobs running thousands or even millions of core hours, or applying self-adaptive methods such that the applications adjust (and thus optimize) themselves for the underlying architecture, dynamically, autonomically, during run-time.

In our presentation, we will concentrate on the use of HPC systems, grids and clouds for running scientific and engineering simulations, using the European DEISA project as an example, analyze different HPC loads and their suitability for supercomputers, grids and for clouds, and take a closer look at adaptive algorithms and their benefits for modern computer architectures. 




francescatoni.jpgFrancesca Toni
Department of Computing, Imperial College London 

Presentation title: Argumentative Agents

Argumentation, initially studied in philosophy and law, has been researched extensively in computing in the last decade, especially for inference, decision making and decision support, dialogue, and negotiation. Simply stated, argumentation focuses on interactions where parties plead for and against some conclusion. In its most abstract form, an argumentation framework consists simply of a set of arguments and a binary relation representing the attacks between the arguments.
By instantiating the notion of arguments and the attack relations, different argument systems can be constructed, predominantly based upon logic. Argumentation provides a powerful mechanism for dealing with incomplete, possibly inconsistent information. It is also fundamental for the resolution of  conflicts and differences of opinion amongst different parties. Further, it is useful for  "explaining" to users outcomes generated automatically.

In the talk, I will focus on the use of argumentation to support intelligent agents in multi-agent systems, in general and in the EC-funded FP6 IST project ARGUGRID  (www.argugrid.eu) and the Agreement Technology COST action IC 0801 (http://www.agreement-technologies.eu/) projects. In particular, I will discuss how argumentation can help agents to make decision, either in isolation (by evaluating pros and cons of conflicting decisions)  or in an open and dynamic environment (by assessing the validity of information they become aware of). I will also illustrate how argumentation can support negotiation and conflict resolution during dialogues. Finally, I will show how arguments can improve the assessment of the trustworthiness of agents in contract-regulated interactions.


 Copyright © PTI 2006-2013  ::  Code/Design by T.S.  ::  Redesigned by A.P.  ::  Webmaster 
 Valid XHTML 1.0 Transitional Valid CSS v 2.1 Valid RSS feed 
Total visits: 750688 times.