Oriented soft localized subspace classification

نویسندگان

  • Thiagarajan Balachander
  • Ravi Kothari
چکیده

Subspace methods of pattern recognition form an interesting and popular classiication paradigm. The earliest sub-space method of classiication was the CLass Featuring Information Compression (CLAFIC) which associated with each class a corresponding linear subspace. Local subspace classiication methodologies which have enhanced classii-cation power by associating more than one linear subspace with each class have also been investigated. In this paper we introduce the Oriented Soft Regional Subspace Classiier (OS-RSC). The highlights of this classiication methodology are (i) Class speciic subspaces are formed such that they speciically maximize average projection of one class while minimizing the average projection of the rival class (ii) Multiple manifolds are formed for each class which gives the classiier greater classiication power (iii) a soft sharing of the training patterns again allows for better classiication performance. The performance of the proposed classiier is tested on real-world classiication problems. Also, it turns out that for the cost function under consideration (that forms class speciic subspaces) the maxima is achieved for a subspace of unit dimensionality. This simpliies the clas-siier structure.

برای دانلود متن کامل این مقاله و بیش از 32 میلیون مقاله دیگر ابتدا ثبت نام کنید

ثبت نام

اگر عضو سایت هستید لطفا وارد حساب کاربری خود شوید

منابع مشابه

Object-oriented subspace analysis for airborne hyperspectral remote sensing imagery

An object-oriented mapping approach based on subspace analysis of airborne hyperspectral images was investigated in this paper. Hyperspectral features were extracted based on subspace learning approaches, in order to reduce the redundancy of spectral space and extract the characteristic images for the further object-oriented classification. In this paper, three kinds of spectral feature extract...

متن کامل

Subspace Feature Analysis of Local Manifold Learning for Hyperspectral Remote Sensing Images Classification

Dimensionality reduction and segmentation have been used as methods to reduce the complexity of the representation of hyperspectral remote sensing images. In this study, a new object-oriented mapping approach is proposed based on nonlinear subspace feature analysis of hyperspectral remote sensing images. Nonlinear local manifold learning approaches for feature extraction were utilized to obtain...

متن کامل

USING DISTRIBUTION OF DATA TO ENHANCE PERFORMANCE OF FUZZY CLASSIFICATION SYSTEMS

This paper considers the automatic design of fuzzy rule-basedclassification systems based on labeled data. The classification performance andinterpretability are of major importance in these systems. In this paper, weutilize the distribution of training patterns in decision subspace of each fuzzyrule to improve its initially assigned certainty grade (i.e. rule weight). Ourapproach uses a punish...

متن کامل

Object-Oriented Classification of Hyperspectral Remote Sensing Images Based on Genetic Algorithm and Support Vector Machine

This paper proposes a method of reducing dimensions based on genetic algorithm and object-oriented classification based on support vector machine (SVM). The basic idea is subspace decomposition of hyperspectral images at first, then selecting suitable bands in each subspace by using genetic algorithm and putting all selected bands of each subspace together. Furthermore, the hyperspectral image ...

متن کامل

Generalized Mutual Subspace Based Methods for Image Set Classification

The subspace-based methods are effectively applied to classify sets of feature vectors by modeling them as subspaces. It is, however, difficult to appropriately determine the subspace dimensionality in advance for better performance. For alleviating such issue, we present a generalized mutual subspace method by introducing soft weighting across the basis vectors of the subspace. The bases are e...

متن کامل

ذخیره در منابع من


  با ذخیره ی این منبع در منابع من، دسترسی به آن را برای استفاده های بعدی آسان تر کنید

برای دانلود متن کامل این مقاله و بیش از 32 میلیون مقاله دیگر ابتدا ثبت نام کنید

ثبت نام

اگر عضو سایت هستید لطفا وارد حساب کاربری خود شوید

عنوان ژورنال:

دوره   شماره 

صفحات  -

تاریخ انتشار 1999