With application in fast data mining

Author

Haghighi, Elham Bavafaye ; Rahmati, Mehdi ; Palm, Gunther

Author_Institution

Comput. Eng. & IT Dept., Amirkabir Univ. of Technol., Tehran, Iran

fYear

2013

fDate

28-30 May 2013

Firstpage

219

Lastpage

224

Abstract

Reducing Computational complexity is a major issue in data mining. Mapping to Multidimensional Optimal Regions (M²OR) is a special purposed method for multiclass classification task. It reduces computational complexity in comparison to the other concepts of classifiers. In this paper, the accuracy of M²OR increases using Learning Inductive Riemannian Manifold in Abstract from (LIRMA). LIRMA estimates the underlying structure of a dataset with respect to the embedded dynamical system of data. The estimated non-linear mapping of LIRMA has the advantage of being topology preserving and inductivity. As a result, the optimal regions are determined more precisely. Consequently, the accuracy of M²OR increases and the memory complexity decreases accordingly.

Keywords

data mining; data structures; learning by example; pattern classification; LIRMA; computational complexity reduction; dataset structure; fast data mining; learning inductive Riemannian manifold in abstract form; memory complexity; multiclass classification; multidimensional optimal regions; nonlinear mapping; topology preservation; Abstracts; Accuracy; Classification algorithms; Computational complexity; Data mining; Manifolds; Principal component analysis; Learning Inductive Riemannian Manifold in Abstract form; Non-linear Mapping to Multidimensional Optimal Regions; embedded dynamical system; fast data mining; multi-classifier; non-linear mapping;

fLanguage

English

Publisher

ieee

Conference_Titel

Information and Knowledge Technology (IKT), 2013 5th Conference on

Conference_Location

Shiraz

Print_ISBN

978-1-4673-6489-8

Type

conf

DOI

10.1109/IKT.2013.6620068

Filename

6620068