Research Activities

Bachelor Degree

 

Thesis - Fingerprint Clustering Using Fuzzy Neural Network, 2004.

 

Abstract : The main objective of this project is to build a Fingerprint Clustering System using Fuzzy Logic (FL) as a sub model and combine with Neural Networks (NN) as Fuzzy Neural (FN) Fingerprint Clustering System as whole. The system proposed is an expansion version of an available Fingerprint Clustering System with the purpose to speed up the matching time of previous version. The first chapter of this report will introduce the previous system’s background and the suggestion on how to improve the matching time by applies clustering methodologies. The objective of this project has been discussed too. The second chapter is the literature review about classifier methodologies which consists of the description of all the classification of clustering algorithms. This report also mentioned which type of classifier is suitable for the proposed fingerprint clustering system. Besides, Fuzzy Logic technology, Neural Networks technology and Fuzzy Neural technology in multiple practical application fields and each performance also mentioned in this chapter. Analyze on other applications and results are useful to generate a possible framework for the proposed system. The next chapter is the problem analysis of the proposed system. A detail analysis of the previous system and its problem will discuss in this chapter. Some possible enhancement methods to solve those problems also have been carried out. In the system overview, a chart of the whole system has been described which included the new clustering module. For the next chapter, the proposed clustering methodologies (FL, NN and FN) and also the comparison with similar methodologies have been described in details to solve the problems of proposed fingerprint clustering system. Last but not least, the experiment results and discussion are also take place for us to analyze the system efficiency.   

 

 

Master Degree - current research

 

A Machine Learning Method For Resource Allocation in Multi-Agent System.

 

Abstract : Multi-agent systems, systems that applied various autonomous agents for goal accomplishment, are very useful in the field of resource allocation. However, these systems still need to be extended to design agent having high capability of negotiation skill, which learn through the content of each interactions. One way of doing that is to analyze tactics and strategies proposing by agent in every counter-offer. This paper addresses these issues by presenting an Artificial Intelligent approach to learn the behavior of other agents in the process of negotiation. A simulation of agent negotiation with multiple issues is used to illustrate the proposing approach.