Polishchuk U. Data compression using autoassociative neurolike structures.

Українська версія

Thesis for the degree of Candidate of Sciences (CSc)

State registration number

0411U001513

Thesis Registration Form

0411U001513.pdf

Applicant for

Polishchuk Uliana Volodymyrivna

Specialization

05.13.23 - Системи та засоби штучного інтелекту

Date of defense

15-02-2011

Specialized Academic Board

Д 35.052.14

Lviv Polytechnic National University

Essay

Thesis is dedicated to development of high-performance neural compression methods both without and with stepless loss regulation. In thesis the architecture of data compression system on the basis of neurolike geometrical transformation machine (GTM) structures in the autoassociative training mode with representation the neuron's output signals in a fixed-point number format is developed. Methods of additional increasing the effectiveness of neural data compression by the example of images are implemented. The two-step compression method by successive use the GTM network and lossless algorithms is development and approved for the first time. The perfected image compression method by division them into frames for GTM network, which provides additional increase the compression ratio, is developed and approved for the first time. The universality of developed neural data compression methods and means is verified by applying them for audio files.

Thesis supervisor

Rashkevych Yuriy Myhajlovych

Official opponents

Романишин Юрій Михайлович
Гече Федір Елемірович

Files

Автореферат--aref.doc

Дисертація Поліщук - версія пороздільно--CCF11012011_00000.jpg

Дисертація Поліщук - версія пороздільно--CCF20012011_00000.jpg

Дисертація Поліщук - версія пороздільно--CПИСОК ВИКОРИСТАНИХ ДЖЕРЕЛ.doc

Дисертація Поліщук - версія пороздільно--ВСТУП.doc

Дисертація Поліщук - версія пороздільно--ДОДАТОК А.doc

Дисертація Поліщук - версія пороздільно--РОЗДІЛ 1.doc

Дисертація Поліщук - версія пороздільно--РОЗДІЛ 2.doc

Дисертація Поліщук - версія пороздільно--РОЗДІЛ 3.doc

Дисертація Поліщук - версія пороздільно--РОЗДІЛ 4.doc

Similar theses

0523U100039

Melnykova Natalya I.

Models and methods for supporting personalized solutions in medical systems.

0421U102573

Vlasenko Oleksandr Mykolayovych

Soft Computing methods and models of risk analysis processes intellectualization

0521U101170

Povkhan Ihor Fedorovych

Methods and principles of constructing discrete object classification trees for data mining.

0521U100926

Pavlyshenko Pavlyshenko Mykhaylovych

Methods of intellectual analysis of consolidated data for decision-making support

0421U101008

Mishchuk Oleksandra Sergijivna

Neural-like methods and tools to forecast the parameters of atmospheric air pollution