Aisimam - An Artificial immune system based intelligent multiangent model

Show full item record

Redirect: RIT Scholars content from RIT Digital Media Library has moved from to RIT Scholar Works, please update your feeds & links!
Title: Aisimam - An Artificial immune system based intelligent multiangent model
Author: Sathyanath, Srividhya
Abstract: The goal of this thesis is to develop a biological model for multiagent systems. This thesis explores artificial immune systems, a novel evolutionary paradigm based on the immunological principles. Artificial Immune systems (AIS) are found to be powerful to solve complex computational tasks. The main focus of the thesis is to develop a generic mathematical model that uses the principles of the human immune system in multiagent systems (MAS). The components and properties of the human immune system are studied. On understanding the concepts of A/5, a literature survey of multiagent systems is performed to understand and compare the multiagent concepts and AIS concepts. An analogy between the immune system parameters and the agent theory was derived. Then, an intelligent multiagent model named AISIMAM is derived. It exploits several properties and features of the immune system in multiagent systems. In other words, the intelligence of the immune systems to kill the antigen and the characteristics of the agents are combined in the model. The model is expressed in terms of mathematical expressions. The model is applied to a specific application namely the mine detection and defusion. The simulations are done in MATLAB that runs on a PC. The experimental results of AISIMAM applied to the mine detection problem are discussed. The results are successful and shows that AISIMAM could be an alternative solution to agent based problems. Artificial Immune System is also applied to a pattern recognition problem. The problem experimented is a color image classification problem useful in a real time industrial application. The images are those of wooden components that need to be classified according to the color and type of wood. To solve the classification task, a simple negative selection and genetic algorithm based A/5 algorithm was developed and simulated. The results are compared with the radial basis function approach applied to the same set of input images.
Record URI:
Date: 2002-11

Files in this item

Files Size Format View
SSathyanathThesis11-2002.pdf 18.84Mb PDF View/Open

The following license files are associated with this item:

This item appears in the following Collection(s)

Show full item record

Search RIT DML

Advanced Search