1 AIT Asian Institute of Technology

Configuring neural networks as iterated function systems

AuthorShaukat, S. Muhammad Shamshad
Call NumberAIT Thesis no. CS-92-23
Subject(s)Neural networks (Computer science)

NoteA thesis submitted in partial fulfillment of the requirements for the degree of Master of Engineering, School of Engineering and Technology
PublisherAsian Institute of Technology
AbstractThe purpose of this study is to configure a neural network as iterated function systems for achieving high image data compression rate at a very high speed. This study comprises three parts. The first part of the study is based on iterated function systems (IFSs) and the usefulness of Collage theorem in finding the IFS codes. The second part shows how random/orbit iteration algorithms can be used to decode or generate the figure back from its IFS encoding. Since both encoding and decoding are computational intensive, so the real application of the method is dependent on fast parallel methods of doing it. In the last section, neural networks are explored as fast, concurrent and distributed mechanism to behave like IFSs. As a result, a high compression ratio of 10,000: 1 can be achievable at a speed suitable for real time simulation of animated sequence of pictures.
Year1992
TypeThesis
SchoolSchool of Engineering and Technology (SET)
DepartmentDepartment of Information and Communications Technologies (DICT)
Academic Program/FoSComputer Science (CS)
Chairperson(s)Sadananda, Ramakoti ;Huynh, Ngoc Phien
Examination Committee(s)Hosomura, Tsukasa
Scholarship Donor(s)Government of Norway (NORAD) ;
DegreeThesis (M.Eng.) - Asian Institute of Technology, 1992


Usage Metrics
View Detail0
Read PDF0
Download PDF0