Pre-conference Workshops/Tutorials

Half-day and full-day pre-conference workshops and tutorials were offered on Monday 2 July 2007.

The pre-conference workshops and tutorials are optional activities at an additional cost and are not included in the Complex'07 Conference registration fee. Participants must be registered delegates to attend the pre-conference workshops and tutorials.

Pre-conference workshop or tutorial registration fees:

  • Standard full-day or half-day registration - $A150
  • Fulltime student registration - A$60.

The pre-conference workshop or tutorial registration fee includes morning/afternoon tea and a light lunch. Half-day workshop or tutorial participants receive morning or afternoon tea and a light lunch. Handouts may be provided for the relevant session only.

Participants are required to choose their workshop or tutorial by marking the relevant section on the conference registration form. Numbers are strictly limited. The Program Organising Committee reserves the right to cancel or amend workshops and tutorials if minimum numbers are not reached.

Tutorial 1

Non-linear dynamics: From chaos to fractals to fractional calculus - Bruce Henry (University of New South Wales). Half day - morning.

This tutorial will provide an overview of dynamical systems including stability theory, bifurcations, chaos, and fractals. Methods will be described for measuring Liapunov exponents, Melnikov functions, and Multifractal properties. A short introduction to fractional calculus and its growing importance as a tool in complex systems research will also be provided.

Tutorial 2

Introduction to modelling and simulation - Robert Marks (University of New South Wales). Half day - morning

see www.agsm.edu.au/~bobm/teaching/C07WS.html

Simulation of social interactions focuses on the patterns that emerge in social systems of interacting agents. "Emergent" means not simply the agregation of individual agent's actions. To understand such "complex adaptive" systems, artificial societies composed of interacting adaptive agents can be created and analyzed. Such models can exhibit properties such as cooperation, social norms, and social stratification, and help understanding such phenomena. Using simulation, previously inaccessible questions are amenable to analysis. Such bottom-up simulations can also be used to design the rules of engagement, for instance in a new market. The workshop will cover theory and practice, with hands-on simulations.

Tutorial 3

Systems biology - Jesus Lopez, Peng Wen and Anke Lehmensiek (University of Southern Queensland). Half day - afternoon.

We regret to advise that due to illness the systems biology tutorial was cancelled.

Tutorial 4

How might biological learning relate to machine learning? - Anthony Bell (University of California). Half day - afternoon.

This tutorial will review results in probabilistic machine learning, and studies of plasticity in the nervous system. The goal is to show exactly what hurdles must be overcome so that the two may be said to be talking about the same thing. Issues include: Unsupervised learning methods such as Independent Component Analysis and extensions thereof, and their limitations, reward-based learning schemes, problems in sensory-motor learning, the relation of learning to Bayesian inference methods, learning with spikes, synaptic physiology, a nascent theory of 'learning across levels', and what this means for functionalism. The arguments presented from an alternative to mainstream positions on these topics.

Workshop 1

Understanding complex networks: Annual meeting of the Network Theory Working Group - David Newth (CSIRO). Full day.

We regret to advise that this workshop was cancelled.

Workshop 2

Introduction to agent-based modelling using NetLogo - Luis Izquierdo (Universidad de Burgos). Full day.

Attendees at this tutorial will start by learning the fundamentals of agent-based modelling. Subsequently, they will be taught the basics of NetLogo (Wilensky, 1999), "what at the time of writing is the best of the agent-based simulation environments" (Gilbert, 2007). Everyone in this tutorial will significantly enhance their knowledge of agent-based modelling using NetLogo, no matter their prior level of expertise. The tutorial will be very practical and interactive. In particular, by the end of the tutorial attendees will have implemented their own agent-based model, even if they have no previous programming experience or knowledge of complex systems.

Workshop 3

Which ontology and heuristics for agent-based modelling? - Pascal Perez (Australian National University) and Senthold Asseng (CSIRO). Half day - morning.

Artificial Intelligence states that agents are inherently rational and intentional entities. Based on these two axiomatic conditions, agents develop cognitive capacities based on formal logic inferences. Though internally consistent, these assumptions often face two limitations when confronted with reality. First, facts don't always support human rationality and intentionality. Second, even when these assumptions seem plausible, formal logic inferences prove incredibly hard to validate against empirical evidence.
These two limitations, associated with the strenuous task of developing the corresponding algorithms, have driven a large number of Agent-Based Modellers away from the AI paradigm. Instead, they have developed empirical heuristics to describe limited cognitive processes. These heuristics are often derived from ethnographic or sociological surveys and aim at describing a specific set of behaviors in a given environment. The drawback of such approaches has fueled most of criticisms against ABM and might be summarized as: "as many models, as many heuristics!"
The proposed workshop will explore the outcomes of a modelling experiment that will take place the week before Complex'07. During this experiment, conducted by our colleague Senthold Asseng (CSIRO-PI, Perth, WA) three international experts will design, explain, and demonstrate to a panel of scientists and managers, different models representing a same reality. These three visions will rely on different heuristics, and eventually on different ontologies, to describe human-landscape relationships.
We invite researchers interested in cognitive and behavioral models, social simulations, and artificial societies to attend this workshop in order not only to learn from the planned experiment but also to contribute to a collective reflection on:
  • How can simple heuristics describe complex realities?
  • Can different ontologies efficiently describe a same reality?
  • Is there a way forward towards generic ontologies and reusable heuristics?
This session is sponsored by the COSNet-ABM theme, CSIRO's Agent Based Modelling Group (CABM), and HEMA Consulting Pty Ltd.

Workshop 4

Engineering complex software-intensive systems - Composition trees and behavior trees - Geoff Dromey (Griffith University), Dan Powell (Griffith University), Shaun Wilson (Aerospace Concepts). Half day - morning.

Systems Integration companies increasingly face the challenge of developing software-intensive systems of unprecedented scale and complexity. Mainstream systems and software engineering has not convincingly won the battle against project wastage that frequently accompanies such projects. The core problem is how to develop the 'right' system when there are large numbers of requirements (hundreds or even thousands), the requirements contain many unknown defects and deficiencies, and it is necessary to deploy a large team to complete the project within a specified time-frame and/or for a fixed cost. Representations that avoid causing short-term memory overflow and produce integrated views that enable high-yield early defect detection are central to making real progress with this core problem - Behavior Trees and Composition Trees do this. They make maximum use of requirements information by supporting rigorous translation and composition of individual requirements one at a time into two integrated graphical views. The justification for, the processes, representations and results of applying this constructive approach to tackling the core problem hindering productively developing large-scale, dependable systems will be presented.

Workshop 5

Social Network Analysis (SNA) and the study of emergence in complex networks - Malcolm Alexander (Griffith University). Half day - afternoon.

Social Network Analysis (SNA) is a well developed area within sociology and the social sciences. In recent years natural scientists have entered the field and extended its methods into a new 'science of networks' with applications in biology, ecology and further areas. Reciprocally, SNA has engaged with concepts of emergence and complexity theory and developed procedures for the statistical modelling of network emergence.
This workshop presents the basic template for handling network data developed for SNA visualisation and analysis packages (NetDraw and UCINET). It demonstrates how to prepare data for these packages and how to translate the basic parameters of network structure into defining conditions for network simulation, estimation and goodness of fit available from exponential random graph modelling (ERGM) software (PNet).
The workshop is designed for people researching any type of network to explore how SNA may add value to their research and its presentation.
Presenter: A/Prof Malcolm Alexander is a sociologist who has worked on networks of interlocking corporate directorates in Australia with comparative studies of Canada, the US and Europe. He is a specialist in 2-mode network analysis and is currently writing a book for Sage on the fundamentals of social network analysis.

Disclaimer
The Complex'07 Conference Organising Committee reserves the right to amend the conference and pre-conference workshop/tutorial program. The Conference Organising Committee, Program Committee and the Conference Managers will not accept responsibility for any act or omission of speakers from the program.