BRIMS is a joint venture between the the Isaac Newton Institute for Mathematical Sciences in Cambridge and Hewlett-Packard's European Research Laboratory in Bristol. Further information on BRIMS is available on the World Wide Web at URL:http://www-uk.hpl.hp.com/brims/.
The early promise of new techniques beyond linear systems theory, such as time-delay embedding and neural networks, has been met by the reality that most systems of interest tend to be not just nonlinear but also nonstationary due to external inputs and stochasticity. For these and other reasons, most people continue to rely on linear models.
A new generation of techniques are now emerging that retain the power of the original insights but are much more broadly applicable. We believe that it is time to ask whether such techniques are ready to leave the security of the laboratory and face the complexities of operation in the real world. Our hope is that the world will be revolutionized by a new generation of tools that could make the design and description of much more complex systems feasible, as long as the attendant opportunities for producing entirely new kinds of nonsense can be avoided.
In advance of the workshop we will make available to participants a collection of data sets from varying industrial and academic sources. Each set will be accompanied by a brief indication of its provenance and a list of questions raised by the data provider. We ask that you apply your techniques to some or all of these sets prior to the workshop, to provide a background for the discussion. In addition, a lab will be set up at the workshop for further analysis of both these data sets and some instrumented experimental systems.
The participants will include representatives from data providers as well as scientists working on the range of applicable techniques. We hope that the outcome will be a greatly enhanced understanding of the scope and limitations of the new developments, that it will stimulate new areas of investigation, and that it will enable closer collaborations between scientists working on data analysis and engineers interested in their results.
The workshop will be held at BRIMS in Bristol, England. Participation will be by invitation only. General questions of an organizational nature should be addressed to Roz Rowe on brims@hplb.hpl.hp.com. Scientific and mathematical questions may be addressed to any member of the organising committee.
The organising committee:
Lee Barford, HP Labs Palo Alto, barford@hpl.hp.com
Neil Gershenfeld, MIT, neilg@media.mit.edu
Jeremy Gunawardena, BRIMS, jhcg@hplb.hpl.hp.com
Bernard Silverman, Bristol University, B.W.Silverman@Bristol.ac.uk
Colin Sparrow, Cambridge University, c.sparrow@newton.cam.ac.uk
Nick Tufillaro, Whitman College, nbt@reed.edu