Place: Room 621/622, GP South (Building 78)
Time: Thursday 29th November, 10:30 morning Tea, 11:00am seminar

Title: In-situ Distributed Genetic Programming for On-line Learning in Complex Systems

Speaker: Philip Valencia, CSIRO ICT
Host: Prof Peter Lindsay

Abstract:

In Nature, evolution utilises energy to both optimise the current  olution in the current environment as well as exploring innovative   solutions for future unknown environments. Over many entity lifetimes, multi-agent systems can evolve to exhibit intelligent   collective behaviours such as that seen with bees and ants. The collective behaviour is a result of the co-evolution of the agents' embodiments as well as their on-line adaptation and learning capacity (i.e. brains). Interestingly higher-order systems can evolve logic solutions to many problems within the lifetime of the agent, despite a fixed embodiment.  Such exploration and optimsation of logic can be described as "playing" or "training", while the exploitation of logic is akin to "work". For these concepts to even
exist, a number of specific conditions must be present which will be discussed in this presentation.

This research conjectures that agents within a complex real world environment can perform genetic evolution of logic to yield desirable and adaptive collective system behaviours.  Typically the logic for man-made complex systems is devised off-line, centralised and biased by human intuition and simulation accuracy. It will be
argued that such solutions are inevitably "brittle" and that the in-situ evolution of logic is preferable for specific problem domains.

Biography:

Philip Valencia received his Bachelors degrees in Engineering (Electronic) and Information Technology from QUT, Australia in 2001. He worked at the CSIRO in Sydney researching real-time time facial recognition and machine learning before his involvement in the Ageless Aerospace Vehicle (AAV) project - a NASA funded project
looking at futuristic aerospace vehicles. For the next 3 years Philip's research focussed on Multi-Agent System (MAS) and Complex Systems Science as methods for achieving self-diagnosing and self-repairing structures. Through this Philip became involved in distributed sensing and actuation for environmental resource
management before returning to Brisbane as a research engineer for the newly formed CSIRO ICT Centre (Autonomous Systems Laboratory). Here he continued researched on Wireless Sensor Networks (WSN).

Dr. Ariel Liebman  
Research Fellow
ARC Centre for Complex Systems
School of ITEE, University of Queensland
email: a.liebman@uq.edu.au
tel: +61-7-3365-1623
mobile: +61-(0)414-226-336
room: 78-414, GP South, St Lucia Campus
web: www.accs.edu.au