Emergent Computing - Introduction to the Special Theme

by Heather J. Ruskin and Ray Walshe

Emergent Computing has become more topical in the last few years and has recently been categorised as a research field in its own right. The history of the field lies primarily in Artificial Intelligence, Numerical Methods and Complexity Theory all of which have contributed in no small part to Emergent Computing.

Emergent Computing is sometimes described as 'highly complex processes arising from the cooperation of many simple processes', ie high-level behaviour resulting from low- level interaction of simpler building blocks. One example of emergent behaviour that has been studied over recent years is that of 'flocking'. In a 'flock of birds', the flock is a dynamic entity consisting of hundreds (or thousands) of individuals. The flock constantly changes shape and direction but maintains overall cohesion. Using cellular automata and treating each bird in the flock as an autonomous agent, with simple local rules controlling agent behaviour relative to closest neighbouring birds, a system can be constructed where the overall behaviour of the collective agents reflects the behaviour of the real flock.

The Emergent Computing paradigm both explores and relies upon biologically and socially inspired systems, in which complex behaviour at the global level emerges in a non-linear manner from large numbers of low-level component interactions. Building software systems, using this component methodology, offers many advantages for solving complex problems, since the algorithmic complexity is achieved through software that is simple and flexible compared to conventional software development techniques

Building systems, with behaviour more than the sum of its parts, attracts methodologies and techniques from a number of disciplines, so we include as a broader definition of the topic, the mathematical and computational techniques that underpin the area of Emergent Computing. These include examples such as classifier systems, neural networks, biological immune systems, autocatalytic networks, adaptive game theory, chaos theory, general nonlinear systems and artificial life.

Recent increased interest in these topics is illustrated by the many international conferences and workshops aimed at Emergent Computing, Emergent Properties, Complexity and Co-Evolution, to name but a few. The European Commission has also recently published the fifth call for proposals under the 'information society technologies' (IST) priority of the Sixth Framework Programme (FP6), which falls under the seminal programme of 'integrating and strengthening the European Research Area'. Significantly, a specific targeted research initiative, included in FP6, is the research heading 'Simulating Emergent Properties in Complex Systems'.

Emergent Computing is influenced by and borrows heavily from other disciplines and one of the most prolific symbioses has been that with Biology. Systems Biology has provided many bio-inspired approaches, (Biomimetic models and methods), giving rise to emergent properties. Neural networks, (from the biological 'neuron' operation), which form the basis of many clustering, classification and learning systems, provide one of the earliest examples. Emergence of conditions, favourable to viral invasion and disease spread, has been modelled using cellular automata and autonomous agent methodologies. The behaviour patterns of swarming bees, flocking birds and schools of fish, have generated so-called swarm technologies, used in complex dynamic systems problems such as scheduling, optimisation and space exploration. Similarly, a portfolio of mathematical and numerical methods, statistical and probabilistic reasoning, have found applications in learning systems for adaptation and prediction, linguistics, intelligence and control.

As the field of emergent computing progresses and computational power allows us to look more closely at the individual components, to manipulate them more efficiently, and to combine and compare readily existing techniques with new and hybrid methodologies, evolutionary complexity seems almost within our grasp. As size and sophistication increase of what can be modelled, useful insight may be gained on systems and applications, with scales ranging from the sub-cellular to the social and beyond.

This issue aims to highlight some of the areas and techniques of current interest internationally in Emergent Computing research, through both special theme articles submitted and invited articles from key researchers.
The invited articles introduce topics such as:

• unmanned space exploration using intelligent swarm technologies
• co-evolution of emergent properties in collective systems (collective intelligence)
• evolution of computer programs by artificial selection, using genetic programming, to obtain human-competitive game-playing strategies.

The 21 special theme articles in this issue highlight current research trends, not only within the ERCIM community, but more widely across Europe. Included are topics, such as:

• evolving higher order intelligence and evolving classifier systems
• particle swarm optimisation in microwave backscattering data and metaheuristics in which a decentralized design leads to emergent phenomena
• fuzzy coalitions in game playing
• self-organised and self-optimisation in dynamic networks and chemical programming
• predictive modelling in financial networks and statistical modelling frameworks
• evolvable hardware using a multi-cellular organism approach
• emergent development in the walking behaviour of a robot
• agent based modelling of viral infection and network emergence in the immune system
• collaborative online development of modular intelligent agents and emergent intelligence in competitive multi-agent systems.

Link:
http://bioinformatics.computing.dcu.ie

Please contact:
Heather J. Ruskin
Dublin City University/IUA, Ireland
E-mail: Heather.Ruskincomputing.dcu.ie

Ray Walshe
Dublin City University/IUA, Ireland
E-mail: Ray.Walshecomputing.dcu.ie