Linear Separability Analysis for Stacked Generalization Architecture

2009-04-11
Ozay, Mete
Vural, Fatos T. Yarman
Stacked Generalization algorithm aims to increase the individual classification performances of the classifiers by combining the information obtained from various classifiers in a multilayer architecture by either linear or nonlinear techniques. Performance of the algorithm varies depending on the application domains and the space analyses that affect the classification performances could riot be applied successfully.

Suggestions

Layout optimization of trusses using simulated annealing
Hasançebi, Oğuzhan (2000-09-08)
This paper addresses to the development of a simulated annealing (SA) based solution algorithm which is automated to achieve the simultaneous optimum design of truss type structures with respect to size, shape and topology design variables. The proposed algorithm is designed in such a way that together with applicability to practical design problems, it is also aimed at producing efficient and improved design solutions for the problems of interest. From the practicality point of view, the task is chosen as ...
Development of an incompressible navier-stokes solver with alternating cell direction implicit method on structured and unstructured quadrilateral grids
Baş, Onur; Tuncer, İsmail Hakkı; Department of Aerospace Engineering (2007)
In this research, the Alternating Cell Direction Implicit method is used in temporal discretisation of the incompressible Navier-Stokes equations and compared with the well known and widely used Point Gauss Seidel scheme on structured and quadrilateral unstructured meshes. A two dimensional, laminar and incompressible Navier-Stokes solver is developed for this purpose using the artificial compressibility formulation. The developed solver is used to obtain steady-state solutions with implicit time stepping m...
Shape Optimizations of Metallic Sheets for Improved Near-Zone Scattering, Reflection, and Focusing Characteristics
Altinoklu, Askin; Ergül, Özgür Salih (2017-10-12)
We consider full-wave analysis and optimizations of metallic sheets to obtain desired electromagnetic responses in the near-zone field intensity and power density distributions. Given a metallic sheet of finite size, deformations are applied to achieve diverse scattering and reflection characteristics, particularly focusing and collecting electromagnetic power at different locations. Optimizations are performed by employing genetic algorithms in a multigrid scheme on the discretized geometry, while each tri...
Numerically efficient analysis of slot-lines in multilayer media using closed form Green's functions
Dural Ünver, Mevlüde Gülbin; Aksun, MI (1996-05-16)
A numerically efficient technique for the analysis of slot-line geometries in multilayer media is presented using closed-form Green's functions in the spatial domain employed in conjunction with the Method of Moments (MoM). The computed equivalent magnetic current distribution on the slot is used to determine the power radiated by the slot and the input impedance. In power calculations, the spatial domain Green's functions are approximated as a power series of radial distance, and the integrals involving th...
Unstructured grid generation and a simple triangulation algorithm for arbitrary 2-D geometries using object oriented programming
Karamete, BK; Tokdemir, Turgut; Ger, M (Wiley, 1997-01-30)
This paper describes the logic of a dynamic algorithm for a general 2D Delaunay triangulation of arbitrarily prescribed interior and boundary nodes. The complexity of the geometry is completely arbitrary. The scheme is free of specific restrictions on the input of the geometrical data. The scheme generates triangles whose associated circumcircles contain 'no nodal points except their vertices. There is no predefined limit for the number of points and the boundaries. The direction of generation of the triang...
Citation Formats
M. Ozay and F. T. Y. Vural, “Linear Separability Analysis for Stacked Generalization Architecture,” 2009, p. 786, Accessed: 00, 2020. [Online]. Available: https://hdl.handle.net/11511/64453.