IEEE Symposium Series on Computational Intelligence 2009

2009 IEEE Workshop on Evolvable and Adaptive Hardware

WEAH09 will consist of an a variety of events including:

• An Evolvable and Adaptive Hardware (EAH) tutorial
• Panel discussions on selected topics of interest spanning the field
• Presentations on potential applications from government and commercial entities

WEAH09 will bring together scientists, engineers and students from around the world to examine recent EAH advances, to articulate open issues and to explore opportunities for EAH in new problem domains. A one-track format for paper presentations will maximize participant interaction. The workshop will focus exclusively on issues that cut across the discipline. We invite papers and panels that examine open EAH issues, Papers describing design problems that could be more efficiently solved by replacing existing, non-EAH design methods with new EAH-based design methods are desired. Of particular interest are papers exploring future opportunities for EAH techniques, especially in new problem domains and/or at the systems level. Papers describing incremental improvements to mature EAH problems will not be accepted unless a clear connection is made to the broader EAH issues mentioned above. Potential paper and panel topics include, but are not limited to, the following issues:

Evolvable and Adaptive Hardware Applications
Are there practical problems that require the use of EAH? Of those problems, which can be addressed with today’s technology? Which ones require development of new techniques or technologies? What would those new technologies entail and how would we go about developing them? What is the best way to make systems evolve or adapt with minimal disruption to current operations?

Evolvability of EAH Systems
To be useful, EAH must be able to evolve/adapt efficiently. How does one measure or characterize evolvablity/adaptability of a specific EAH system? What adaptive learning algorithms or evolutionary algorithms are most likely to maximize system evolvablity? Are there any features common across EAH platforms that can be isolated and exploited to improve learning/adaptation of all EAH systems? Are there some EAH platforms that one can expect to be more evolvable than the rest? If so, why?

Understanding EAH Systems
EAH systems can potentially evolve to forms somewhat different than what a human would design. However, in practice one must be able to verify that an evolved or adapted system won’t act unpredictably. How can one verify that an evolved system will be safe as well as effective? Is it wise to let systems decide on their own to begin adapting or is that too dangerous? Are there ways of explaining how the devices function after they are evolved? Are there unforeseen tradeoffs between explainablity and evolvability? If so, what are they?

Future Directions in EAH
What new opportunities are on the horizon? For example, are geographically distributed energy systems that autonomously adapt to handle power failures practical? How can real-time systems use EAH? Are there new material substrates (nanomaterials, quantum systems, etc.) that could be combined with adaptive learning and/or evolutionary computation to create new EAH systems? Can systems predict future failures and prevent them from ever occurring by evolving to some different configuration?

Informal questions should be sent to Andy Tyrrell amt@ohm.york.ac.uk, with the subject heading WEAH09.

Program Chairs:
Andy Tyrrel, University of York, U.K.
Andy Greensted, University of York, U.K.
Garrison Greenwood, Portland State University, USA

Programme Committee:
Arturo Aguirre (Mexico)
John Gallagher (USA)
Garry Greenwood (USA)
Pauline Haddow (Norway)
Didier Keymeulen (USA)
Jason Lohn (USA)
Lukas Sekanina (CZ)
Andre Stauffer (Switzerland)
Adrian Stoica (USA)
Gianluca Tempesti (UK)
Ricardo Zebulum (USA)