ACDM 2006 - CHAIR & SPEAKERS

The Conference will be chaired by Prof Ian Parmee of the University of the West of England, Bristol, UK and will host the ACDM Lecturer and the following Keynote Speakers:

"New Types of Interactive Evolutionary Computing Applications and Latest Research on Reducing User Fatigue"

Dr Zoran Kapelan

Centre for Water Systems

University of Exeter

"Some Recent Developments in Water Distribution Systems Rehabilitation"

Click here for Synopsis

Dr. Evelyne Lutton

Lab Director:

INRIA Saclay - Ile-de-France

http://apis.saclay.inria.fr/

"Overview of a large scale and real-world multi-user IEC application in
e-learning"

Click here for Synopsis

Prof Dave Cliff
Department of Computer Science, University of Bristol, BS8 1UB, UK.

Continuing falls in the real cost of computing power over the last decade have allowed computationally intensive adaptive computing techniques, originally developed in academic and industrial research labs, to be routinely deployed in industrial contexts. The current switch to utility-style provisioning of computing resources offers the possibility of a step-change in what applications are possible, and how those applications are deployed. In this lecture I'll sketch my view of where adaptive computing for design and manufacture (ACDM) is likely to head in the next few years -- this is a view based on several years spent evaluating and deploying ACDM, first for Hewlett-Packard and then for a number of major financial institutions.

Bio:

Dave Cliff is a Professor of Computer Science at the University of Bristol. He previously served in academic faculty jobs at the University of Sussex (UK), at the MIT Artificial Intelligence Lab (USA), and at the University of Southampton (UK). From 1998-2005 Cliff worked as an industrial research scientist: formerly as a Department Scientist at the Hewlett-Packard Labs European Research Centre in Bristol, where he founded and led HP’s Complex Adaptive Systems research group; and latterly as a Director in Deutsche Bank’s Foreign Exchange (FX) Complex Risk Group, where he worked on various aspects of adaptive automation on Deutsche’s City of London FX trading floor. Since October 2005, Cliff has been Director of the UK national research and training initiative in the science and engineering of Large-Scale Complex IT Systems (LSCITS).
Phase One of the LSCITS Initiative commenced in October 2007, has a budget of over £15m, and will involve more than 250 person-years of research effort:: full details are available at www.lscits.org <http://www.lscits.org/>. Cliff is author or co-author on over 70 academic publications, inventor or co-inventor on 15 patents; and he has undertaken advisory and consultancy work for a number of major companies and for the UK Government. He is a visiting professor at the University of Leeds, a chartered fellow of the British Computer Society, a member of EPSRC’s ICT Strategic Advisory Team, and a member of CPHC.

New Types of IEC Applications and Latest Research on Reducing User Fatigue

Hideyuki Takagi
Kyushu University, Japan
http://www.design/kyushu-u.ac.jp/~takagi

The first topic of this talk is to show new types of Interactive Evolutionary Computation (IEC) application researches. Major IEC applications are optimizing target systems and creating graphics, images, shapes, sounds, vibrations, and others. We introduce two new types of IEC applications. The first one is measuring human characteristics.
IEC is an optimization method based on human subjective evaluation.
Likely reverse engineering, we may measure the evaluation characteristics or mental conditions of an IEC user by analyzing the outputs from the target system optimized by the user. The second one is extension of IEC evaluation. Usually IEC optimizes a target system based on IEC user's subjective evaluation, i.e. psychological evaluation. We may extend the evaluation from psychological one to physiological one. We show the framework of the extended IEC.
The second topic of this talk is to overview researches that try to reduce IEC user fatigue and show our latest research in this area.
Several approaches have been proposed to reduce IEC user's fatigue; some of them are improving input/output interface, accelerating EC search, allowing human intervention into EC search, estimating human evaluations, and others. Here, we introduce our latest research and show our view.

This talk starts with presenting a few challenging issues in employing evolutionary algorithms to solve complex design problems, such as how to improve the scalability of evolutionary algorithms to high-dimensional search space, how to reduce the computational cost when quality evaluations are highly time-consuming, and how to enhance robustness of the designs to internal and external perturbations. Existing approaches to addressing these issues will be introduced, focusing on the idea of incorporating domain knowledge into evolutionary algorithms. Two examples will be given to illustrate these ideas. In the first example, we will present a meta-modeling technique for single and multi-objective optimization using multiple surrogates. The second example shows how general domain knowledge, such as regularity in the distribution of Pareto-optimal solutions, can also be taken advantage of to significantly improve the scalability of multi-objective evolutionary algorithms. This talk concludes with some recent idea for engineering design inspired from evolutionary development in biology, where a cell growth model is used for designing of inner structures.

The performance of UK and other water distribution systems (WDSs) is deteriorating with time due to ageing infrastructure and increased urbanisation. At the same time, increased customer and regulatory expectations are to be met. As a consequence, WDSs need to be rehabilitated periodically. Since large costs are typically involved in the process whilst limited budgets are available, the WDS rehabilitation problem is formulated and solved as an optimisation problem. The multiple objectives are used to address the main trade-off between the rehabilitation cost on one side and the system performance on the other side. WDS optimisation problem is a large, complex, non-linear, discrete, computationally demanding problems and hence very difficult to solve. A number of past and recent approaches are presented involving application of a number of advanced optimisation methods including probabilistic model building genetic algorithms. The performance of these algorithms is tested and compared on a number of case studies ranging from benchmark to real-life problems.

Overview of a large scale and real-world multi-user IEC application in e-learning.

Evelyne Lutton
INRIA - Saclay, Faris.
http://apis.saclay.inria.fr/

This talk will give an overview of a multi-user IEC application developed for Paraschool, an important e-leaning company in France,specialising in young childrens' education. An interactive version of a variant of an ACO (ant colony optimisation) technique has been developed, in order to optimise pedagogical paths traversing a set of educational topics organised in a graph. The idea was to create an adaptive interaction in order to propose strength-adapted progressive drills.

New concepts have been introduced (among which elo fitness rating, erosion of pheromones, mixed fitness functions) in order to adapt to the characteristics of the system that involves more than 150000 children.

Real-size experimentations have shown that the system is stable over-time and behaves as expected :
      * The learning of students is improved by an adaptive navigation
        within the Paraschool web pages.
      * The teachers have now access to a new audit tool that detects
        semantic errors among thousands of drills and proposes new
        pedagogical paths.

The system is currently in exploitation by Paraschool (http://www.paraschool.com) and has been developed in collaboration with Pierre Collet (Strasbourg university).