Overview and comparative study of dimensionality reduction techniques for high dimensional data

•Examination of linear and non-linear dimensionality reduction techniques.•Selection of suitable dimensionality reduction techniques for diverse types of data (i.e., text, numeric, signals, etc.).•Investigation of open issues associated with dimensionality reduction techniques in different applicati...

Full description

Saved in:
Bibliographic Details
Published in:Information fusion Vol. 59; pp. 44 - 58
Main Authors: Ayesha, Shaeela, Hanif, Muhammad Kashif, Talib, Ramzan
Format: Journal Article
Language:English
Published: Elsevier B.V 01.07.2020
Subjects:
Dimensionality reduction
Features
High dimensional data
Linear techniques
Nonlinear techniques
Dimensionality reduction
Nonlinear techniques
Linear techniques
Features
High dimensional data
ISSN:1566-2535, 1872-6305
Online Access:Get full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:•Examination of linear and non-linear dimensionality reduction techniques.•Selection of suitable dimensionality reduction techniques for diverse types of data (i.e., text, numeric, signals, etc.).•Investigation of open issues associated with dimensionality reduction techniques in different application.•Exploration of high dimensional data issues and solution using appropriate dimensionality reduction techniques. The recent developments in the modern data collection tools, techniques, and storage capabilities are leading towards huge volume of data. The dimensions of data indicate the number of features that have been measured for each observation. It has become a challenging task to analyze high dimensional data. Different dimensionality reduction techniques are available in literature to eliminate irrelevant and redundant features. Selection of an appropriate dimension reduction technique can help to enhance the processing speed and reduce the time and effort required to extract valuable information. This paper presents the state-of-the art dimensionality reduction techniques and their suitability for different types of data and application areas. Furthermore, the issues of dimensionality reduction techniques have been highlighted that can affect the accuracy and relevance of results.
ISSN:1566-2535
1872-6305
DOI:10.1016/j.inffus.2020.01.005