Using Functional Models to Support Self-Adaptation

Advisor: Prof. Goel

Description

Genetic programming is a powerful method for supporting self-adaptation in software agents. Unfortunately, GP typically amounts to an exhaustive search of the design space. An alternative method for building self-adaptive software agents is to endow them with a meta-model of their own knowledge, reasoning and architecture. I will first read relevant papers and then design a simple software agent (e.g., a fax machine agent, a webpage agent) capable of adapting itself as its requirements evolve.

Motivation

The motivation is primarily for me to learn more about self-adapting systems. Other goals are to potentially open some avenues for future research, and to learn what it is like to work with Prof. Goel.

Communications

Communication with Prof. Goel will be handled partly through email discussions. In addition, I have a class with him three days a week and will talk with him after class or at another mutually convenient time.

Deliverables

A writeup consisting primarily of a design for an adaptive software agent.