Dissertation > Excellent graduate degree dissertation topics show

FSVM Classification Based on Cloud Model

Author: YanXin
Tutor: ZhangYueQin
School: Taiyuan University of Technology
Course: Software Engineering
Keywords: fuzzy support vector machines membership cloud model remote sensing image
CLC: TP181
Type: Master's thesis
Year: 2013
Downloads: 2
Quote: 0
Read: Download Dissertation

Abstract

In today’s information age, with the rapid development of network and computer technology, the amount of information exploding, in order to meet people’s needs, to avoid huge amounts of data in a disorderly state, data is classified as a indispensable means. Fuzzy support vector machine as a very effective classification method can solve many of the traditional classification methods due to samples nonlinear and high dimension that can not be classified or low classification accuracy. In fuzzy support vector machine classification methods, the problem of membership function determining is the most critical research. Because there are a lot of uncertainties in the objective world, may easily lead to the sample of the edge of the classification can not be correctly classified. How to better express the uncertainty of things and phenomena in the objective world is the focus and hot spots of current research in the field of natural science. In the conversion between qualitative and quantitative, the qualitative concept with classification has a variety of uncertainties. Cloud model as a uncertainty model of the conversion of qualitative concept and quantitative numerical can better reflect this uncertainty of the classification concept.To solve the above problems, using the unique principle of qualitative and quantitative conversion of the cloud model and the advantages of nonlinearity, anti-noise capability of fuzzy support vector machine, combining cloud model with fuzzy support vector machine to improve the membership of fuzzy support vector machine. And because the remote sensing image data has the characteristics of boundary ambiguity and interpret process uncertainty, in this paper we use remote sensing images as the experimental data to verify the effectiveness of the improved algorithm and improve the classification accuracy of remote sensing image classification in the set of remote sensing data.The papers main contents are as follows:(1)Describing the research background, research status and significance of the article and analyzing this topic ideas.(2)Introducing the theoretical basis and classification principle of support vector machines, and focusing on the multi-class classification method of support vector machines. Analyzing the fuzzy support vector machine classification method on the basis of the fuzzy set theory, and introducing its core module membership function.(3)Overview of the cloud model, introducing the definition and numerical characteristics of the cloud model and summarizing the characteristics of the cloud model, and finally focusing on three types of cloud model generator.(4)As the remote sensing satellite images for the experimental data, using the distance membership algorithm and fuzzy support vector machine theory to constructed sub-optimal classification hyper-plane. To the generated mixed division and drain sub-samples using the improved membership of the cloud model of the FSVM method to construct the optimal hyper-plane and integrating to achieve classification of remote sensing data, and analyzing classification performance.(5)In order to verify the classification of the improved algorithm, we use the remote sensing aerial image as data sets, and using the same experimental method to classify for comparison, classification performance analysis concluded that the proposed method of remote sensing image classification experiment is feasible, in particular, for the remote sensing satellite images, the classification results is more apparent.

Related Dissertations

CLC: > Industrial Technology > Automation technology,computer technology > Automated basic theory > Artificial intelligence theory > Automated reasoning,machine learning
© 2012 www.DissertationTopic.Net Mobile