Show/Hide Menu
Hide/Show Apps
Logout
Türkçe
Türkçe
Search
Search
Login
Login
OpenMETU
OpenMETU
About
About
Open Science Policy
Open Science Policy
Open Access Guideline
Open Access Guideline
Postgraduate Thesis Guideline
Postgraduate Thesis Guideline
Communities & Collections
Communities & Collections
Help
Help
Frequently Asked Questions
Frequently Asked Questions
Guides
Guides
Thesis submission
Thesis submission
MS without thesis term project submission
MS without thesis term project submission
Publication submission with DOI
Publication submission with DOI
Publication submission
Publication submission
Supporting Information
Supporting Information
General Information
General Information
Copyright, Embargo and License
Copyright, Embargo and License
Contact us
Contact us
Classification through incremental max-min separability
Date
2011-05-01
Author
Bagirov, Adil M.
Ugon, Julien
Webb, Dean
Karasözen, Bülent
Metadata
Show full item record
This work is licensed under a
Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License
.
Item Usage Stats
84
views
0
downloads
Cite This
Piecewise linear functions can be used to approximate non-linear decision boundaries between pattern classes. Piecewise linear boundaries are known to provide efficient real-time classifiers. However, they require a long training time. Finding piecewise linear boundaries between sets is a difficult optimization problem. Most approaches use heuristics to avoid solving this problem, which may lead to suboptimal piecewise linear boundaries. In this paper, we propose an algorithm for globally training hyperplanes using an incremental approach. Such an approach allows one to find a near global minimizer of the classification error function and to compute as few hyperplanes as needed for separating sets. We apply this algorithm for solving supervised data classification problems and report the results of numerical experiments on real-world data sets. These results demonstrate that the new algorithm requires a reasonable training time and its test set accuracy is consistently good on most data sets compared with mainstream classifiers.
Subject Keywords
Classification
,
Data mining
,
Data analysis
,
Supervised learning
,
Piecewise linear classifier
URI
https://hdl.handle.net/11511/29970
Journal
PATTERN ANALYSIS AND APPLICATIONS
DOI
https://doi.org/10.1007/s10044-010-0191-9
Collections
Graduate School of Applied Mathematics, Article
Suggestions
OpenMETU
Core
DETERMINATION OF THE RELAXATION SPECTRUM FROM OSCILLATORY SHEAR DATA
ORBEY, N; DEALY, JM (1991-08-01)
In order to use either a linear or nonlinear model of viscoelasticity to calculate the stress response of a material to various deformations, it is usually necessary to have available an explicit equation for the linear relaxation modulus G(t). The most popular procedure is to use the data from a small-amplitude oscillatory shear experiment to determine the parameters of a generalized Maxwell model. However, this is an ill-posed problem and is not at all a straightforward curve-fitting operation. We comp...
Almost periodic solutions of the linear differential equation with piecewise constant argument
Akhmet, Marat (2009-10-01)
The paper is concerned with the existence and stability of almost periodic solutions of linear systems with piecewise constant argument where t∈R, x ∈ Rn [·] is the greatest integer function. The Wexler inequality [1]-[4] for the Cauchy's matrix is used. The results can be easily extended for the quasilinear case. A new technique of investigation of equations with piecewise argument, based on an integral representation formula, is proposed. Copyright © 2009 Watam Press.
NORMAL SOLVABILITY OF ELLIPTIC BOUNDARY-VALUE-PROBLEMS ON ASYMPTOTICALLY FLAT MANIFOLDS
ERKIP, AK; SCHROHE, E (Elsevier BV, 1992-10-01)
Normal solvability is shown for a class of boundary value problems on Riemannian manifolds with noncompact boundary using a concept of weighted pseudodifferential operators and weighted Sobolev spaces together with Lopatinski-Shapiro type boundary conditions. An essential step is to show that the standard normal derivative defined in terms of the Riemannian metric is in fact a weighted pseudodifferential operator of the considered class provided the metric is compatible with the symbols.
FORCED HARMONIC RESPONSE ANALYSIS OF NONLINEAR STRUCTURES USING DESCRIBING FUNCTIONS
TANRIKULU, O; KURAN, B; Özgüven, Hasan Nevzat; IMREGUN, M (1993-07-01)
The dynamic response of multiple-degree-of-freedom nonlinear structures is usually determined by numerical integration of the equations of motion, an approach which is computationally very expensive for steady-state response analysis of large structures. In this paper, an alternative semianalytical quasilinear method based on the describing function formulation is proposed for the harmonic response analysis of structures with symmetrical nonlinearities. The equations of motion are converted to a set of nonl...
Transformation-based metamaterials to eliminate the staircasing error in the finite difference time domain method
Ozgun, Ozlem; Kuzuoğlu, Mustafa (Wiley, 2012-07-01)
A coordinate transformation technique is introduced for the finite difference time domain method to alleviate the effects of errors introduced by the staircasing approximation of curved geometries that do not conform to a Cartesian grid. An anisotropic metamaterial region, which is adapted to the Cartesian grid and designed by the coordinate transformation technique, is constructed around the curved boundary of the object, and the region occupied between the curved boundary and the inner boundary of the ani...
Citation Formats
IEEE
ACM
APA
CHICAGO
MLA
BibTeX
A. M. Bagirov, J. Ugon, D. Webb, and B. Karasözen, “Classification through incremental max-min separability,”
PATTERN ANALYSIS AND APPLICATIONS
, pp. 165–174, 2011, Accessed: 00, 2020. [Online]. Available: https://hdl.handle.net/11511/29970.