Use this URL to cite or link to this record in EThOS:
Title: Rule discovery from swarm systems
Author: Stoops, David
ISNI:       0000 0004 2715 486X
Awarding Body: University of Ulster
Current Institution: Ulster University
Date of Award: 2011
Availability of Full Text:
Access from EThOS:
Rules determine how a system operates within the boundaries of its given environment. This principle is present within both natural and man-made systems alike. The term "man-made system" refers to systems which have been developed, implementing a specifically designed rule-set for its functionality. Whereas "natural systems" are those rules presented and seen within various species or environments within the natural world. The research presented within this thesis looks at the methods and results of discovering rules from swarm systems. The swarm is a phenomenon seen within the natural world where same species individuals group together to accomplish a task using their behaviours. This has many aspects of difficulty to it, various species employ differing rules, and a species can employ individuals with differing rule-sets. The primary swarm to be investigated is the Bird flock, with a developed simulation being used as a test-bed. The BOlD simulation presents a computer program incorporating a visual representation of the birds, each employing 3 basic rules. The challenge for this work is to develop a method of providing a generic rule discovery system for swarms. This encompasses the ability to collect and manipulate data, and discover the rules present within differing species and roles within swarms. An additional objective is to present a novel method for verifying or validating the rules, employing performance testing. This type of testing provides much insight into the validity of the rules by determining how well they perform when tested against the original rule-set. The data collected from the BOlD simulation was transformed into the selected attributes for mining, with a rule algorithm being used to discover rules. The rules which were discovered provided an insight into the behaviours of the swarm, with the performance testing proving that they were not as effective or efficient as the original rules employed within the simulation. The presented research provides the ability to gather a further insight into swarm behaviours, and presents a varying method for understanding the interactions we view every day within these swarms.
Supervisor: Not available Sponsor: Not available
Qualification Name: Thesis (Ph.D.) Qualification Level: Doctoral
EThOS ID:  DOI: Not available