Invariant pattern recognition using SVDD-based associative memories

Author

Ciocoiu, Iulian B.

Author_Institution

Fac. of Electron., Telecommun. & Inf. Technol., Gheorghe Asachi Tech. Univ. of Iasi, Iasi, Romania

fYear

2013

fDate

11-12 July 2013

Firstpage

1

Lastpage

4

Abstract

Pattern recognition performances of a special gradient-type dynamical system are investigated. The system exhibits stable equilibrium points whose positions are defined by the minima of a data-dependent Lyapunov function constructed using the Support Vector Data Description (SVDD) algorithm. Invariance to standard geometric transformations is inferred by combining SVDD with the tangent distance (TD), which has superior recognition performances when compared to the Euclidean distance. Experimental results using the USPS handwritten characters database and the Olivetti face images database confirm the superiority of the proposed approach over existing solutions.

Keywords

Lyapunov methods; content-addressable storage; data description; face recognition; handwritten character recognition; support vector machines; visual databases; Olivetti face image database; SVDD; SVDD-based associative memories; TD; USPS handwritten character database; data-dependent Lyapunov function; gradient-type dynamical system; invariant pattern recognition performance; stable equilibrium points; standard geometric transformations; support vector data description algorithm; tangent distance; Associative memory; Databases; Pattern recognition; Principal component analysis; Support vector machines; Training; Vectors;

fLanguage

English

Publisher

ieee

Conference_Titel

Signals, Circuits and Systems (ISSCS), 2013 International Symposium on

Conference_Location

Iasi

Print_ISBN

978-1-4799-3193-4

Type

conf

DOI

10.1109/ISSCS.2013.6651250

Filename

6651250