Evolving defense/attack maneuvers in a non-stationary environment
Most real life situations are non-stationary; that is, the objects in the environment are moving and the situation continuously changes. For example, in a battle, the circumstances change over time and fixing a plane before hand is simply non-sense. A plan for attack may need to change its underlying strategy during the battle to adapt to new changes in the environment and may even change to a defense plan. Therefore, building the optimal plan (if it exists initially) for a team does not mean that the team will necessarily win and does not mean that the plan will remain optimal during the battle.
The objective of this project is to use a combination of Darwinian and Lamarkian evolutions (Evolutionary Algorithms + local search) to evolve maneuver strategies that are capable to adopt in a non-stationary environment. Students who are interested to work on this project, are very much encouraged to contact the investigator to discuss the project further.
Emergence of maneuver strategies in a reactive multi-agent system
Traditional cognitive agent approaches are based on hand-wiring the strategies of each agent. The objective of this project is to start with simple strategies combined with self-organization (positive and negative feedbacks, amplification of fluctuations, and stigmergy) to study how complex strategies evolve from simple interactions and domain knowledge. In this project, we will use tools from Bayesian inference and Neural Networks to control the agents. There is also high potentials to use Rule-extraction from Artificial Neural Networks to analyze the underlying knowledge. Students who are interested to work on this project, are very much encouraged to contact the investigator to discuss the project further.
Self-organization in a non-stationary environment of competing agents
Biological systems in a non-stationary environment never reach equilibrium. The environmental pressures always affect the landscape of these systems and therefore a mean for quick and efficient self-organization is important for these systems to survive. In this project, we will use the same concepts to investigate different methods for self-organization and how they work in a non-stationary environment with competing agents. Students who are interested to work on this project, are very much encouraged to contact the investigator to discuss the project further.
Marriage in Honey-bees for evolving strategies in a non-stationary environment
Marriage in Honey-bees is a recent approach developed by the project investigator for planning and scheduling problems represented in propositional satisfiability. The aim of this project is to extend the approach to evolve strategies in a battlefield. Students who are interested to work on this project, are very much encouraged to contact the investigator to discuss the project further.