Sitemap | Contact | 中文 | HFCAS | CAS
Search: 
 
Location: Home > News > Events
IIM Made Significant Progresses in Pattern Classification
Author: WU Qianshen
Update time: 2016-07-21
Close
Text Size: A A A
Print

Classifier design is a fundamental problem in artificial intelligence, machine learning, pattern recognition, and data mining. A lot of pattern classification methods such as the nearest neighbor classifier (NN), sparse representation-based classification (SRC), and support vector machine (SVM) have been proposed in the literature. These typical and widely used classifiers were originally developed from different theory or application motivations and they are conventionally treated as independent and specific solutions for pattern classification.

Recently, the associate professor Jie GUI of Institute of Intelligent Machines (IIM), Hefei Institutes of Physical Science, has made important progresses in pattern classification and the corresponding paper is published in IEEE Transactions on Cybernetics: http://ieeexplore.ieee.org/xpl/RecentIssue.jsp?punumber=6221036.

This paper proposes a novel pattern classification framework, namely, representative vector machines (or RVMs for short) to solve the problem. The basic idea of RVMs is to assign the class label of a test example according to its nearest representative vector. The contributions of RVMs are twofold. On one hand, the proposed RVMs establish a unified framework of classical classifiers because NN, SVM and SRC can be interpreted as the special cases of RVMs with different definitions of representative vectors. Thus, the underlying relationship among a number of classical classifiers is revealed for better understanding of pattern classification. On the other hand, novel and advanced classifiers are inspired in the framework of RVMs. In this paper, a robust pattern classification algorithm named discriminant vector machine (DVM) is motivated from RVMs. Given a test example, DVM first finds its k-nearest neighbors and then performs classification based on the robust M-estimator and manifold regularization. Extensive experimental evaluations on a variety of visual recognition tasks such as face recognition (Yale and FRGC databases), object categorization (Caltech-101 dataset) and action recognition (ASLAN) demonstrate the advantages of DVM over other classifiers.

This work was supported by the grant of the National Basic Research Program of China, the National Science Foundation of China, etc.

Fig. Illustration of the basic ideas of three pattern classifiers. (a) Nearest feature line classifier. (b) Nearest feature plane classifier. (c) Support vector machines.

Image by GUI Jie

Keywords: Classifier; pattern classification; pattern classification framework; representative vector machines; discriminant vector machine

Title of this paper : Representative Vector Machines: A Unified Framework for Classical Classifiers

The link of this paper :

http://ieeexplore.ieee.org/stamp/stamp.jsp?arnumber=7194763

Contact

Prof. GUI Jie, Ph.D Principal Investigator

Institute of Intelligent MachinesChinese Academy of Sciences

(http://www.iim.cas.cn/)

Hefei, Anhui 230000, China

Tel:0551-65590632

E-mail:guijie@ustc.edu

Copyright @2011 Institute of Intelligent Machines, CAS. All Rights Reserved
Email: iim@iim.ac.cn Tel: 0086-551-65591136 Fax: 0086-551-65592420