ALAR
The Artificial Life and Adaptive Robotics Laboratory

 

Name:

Dr. James Michael Whitacre

Research Associate

Qualifications:

PhD, Chemical Engineering, University of New South Wales, Sydney Australia, 2007

BSE, Chemical Engineering, Princeton University, Princeton, NJ, USA, 2003

Publications:
  • Whitacre, J. M., Sarker, R. A., and Pham, T. Q., “The Self-Organization of Interaction Networks for Nature-Inspired Optimization.” IEEE Transactions on Evolutionary Computation, (Accepted March, 2007)  http://www.ceic.unsw.edu.au/staff/Tuan_Pham/Whitacre_SOTEA_2007.pdf  
  • Whitacre, J. M., Sarker, R. A., and Pham, T. Q., “The influence of population topology and historical coupling on Evolutionary Algorithm population dynamics.” Applied Soft Computing, (Submitted September, 2007)
  • Whitacre, J. M., Sarker, R. A., and Pham, T. Q., “A Self-Organizing Topology for distributed Evolutionary Algorithms based on fitness-driven community structures.” IEEE Transactions on Evolutionary Computation, (Submitted September, 2007)
  • Whitacre, J. M., Pham, T. Q., and Sarker, R. A., “Use of statistical outlier detection method in adaptive evolutionary algorithms.” In Proceedings of the 8th Annual Conference on Genetic and Evolutionary Computation (Seattle, Washington, USA, July 08 - 12, 2006). GECCO '06. ACM Press, New York, NY, 1345-1352, 2006. www.ceic.unsw.edu.au/staff/Tuan_Pham/fp122-whitacre.pdf 
  • Whitacre, J. M., Pham, T. Q., and Sarker, R. A., “Credit assignment in adaptive evolutionary algorithms.” In Proceedings of the 8th Annual Conference on Genetic and Evolutionary Computation (Seattle, Washington, USA, July 08 - 12, 2006). GECCO '06. ACM Press, New York, NY, 1353-1360, 2006. www.ceic.unsw.edu.au/staff/Tuan_Pham/fp123-whitacre.pdf

Research Topic:

Scenario Generation Algorithms for Defence Logistics

Research Summary:

In many systems of interest (e.g. global environment, politics, war on terror, financial markets), there is great uncertainty in what the future might hold.  Making informed decisions in such environments is challenging since any expectations of the future are unlikely to become a reality.  For systems which display a sensitivity to initial conditions and an even stronger sensitivity to unexpected events, it is desirable to instead devise strategies which are robust to those events and can adapt to new circumstances.   Many optimization problems of practical interest such as logistics problems actually display these forms of dynamical uncertainty, however there has been little effort to address these problem characteristics (in academia or elsewhere).  For such problems, the desired solution is one that involves an adaptive strategy that recognizes and embraces the uncertainty of future environments.  Evolutionary Computation techniques have a strong potential for addressing these challenges in ways that other techniques (e.g. Stochastic Programming) can not.  

However, a prerequisite to such research efforts is the development of scenario generation algorithms which are able to simulate the dynamics of the real systems and generate adequate sampling of the future possibility space.  Such simulations must be able to adequately model the coupling of component properties, the importance of local interactions, the existence of heterogeneous systems, and the existence of multi-scale systems while at the same time seamlessly integrating expert knowledge about the problem domain and maintaining an intuitive, extensible framework that can be quickly modified for new conditions.  In addition to all these requirements, we are rarely confident that the initially formed model actually resembles the real system and so we also need a model that can evolve over time to better mimic the real system’s dynamical coupling and structure.  My research is focused on the development of new models of complex adaptive systems (e.g. extensions to adaptive networks and agent based models) which can address many of these challenges.

Personal Link: