Multi-objective Optimization for Clustering using Evolutionary Computing
(2014) FMS820 20142Mathematical Statistics
- Abstract (Swedish)
- Clustering of data points is a classic problem that so far has no perfect general
solution. Most existing solutions are computational expensive or simplied due to the combinatorial aspect of the problem. Genetic algorithms are becoming more and more popular due to their ability to solve large scale optimization problems. This thesis investigates the idea of creating a genetic algorithm able to perform clustering, which can be viewed as an optimization problem.
The aim of this thesis was to make the genetic clustering algorithm as general and automatic as possible. The end result is an algorithm that takes as input a data set and an approximate value of the number of clusters wanted, and as output gives a clustering. The algorithm is able... (More) - Clustering of data points is a classic problem that so far has no perfect general
solution. Most existing solutions are computational expensive or simplied due to the combinatorial aspect of the problem. Genetic algorithms are becoming more and more popular due to their ability to solve large scale optimization problems. This thesis investigates the idea of creating a genetic algorithm able to perform clustering, which can be viewed as an optimization problem.
The aim of this thesis was to make the genetic clustering algorithm as general and automatic as possible. The end result is an algorithm that takes as input a data set and an approximate value of the number of clusters wanted, and as output gives a clustering. The algorithm is able to handle categorical values, missing samples and is also able to weight the variables in order to decide which ones that should have the greatest impact on the clustering.
The algorithm is tested on 2- and 4-dimensional data sets with clusters of different shapes and sizes. All tests made, with focus on one or many of the algorithms different features, gave results showing reasonable clustering solutions.
The use of a genetic algorithm to perform clustering seems very promising due to the algorithms ability to nd quality clusters fullling multiple objectives. (Less)
Please use this url to cite or link to this publication:
http://lup.lub.lu.se/student-papers/record/4690510
- author
- Karlsson, Louise
- supervisor
- organization
- course
- FMS820 20142
- year
- 2014
- type
- H2 - Master's Degree (Two Years)
- subject
- language
- English
- id
- 4690510
- date added to LUP
- 2014-09-30 11:37:59
- date last changed
- 2014-09-30 11:37:59
@misc{4690510,
abstract = {{Clustering of data points is a classic problem that so far has no perfect general
solution. Most existing solutions are computational expensive or simplied due to the combinatorial aspect of the problem. Genetic algorithms are becoming more and more popular due to their ability to solve large scale optimization problems. This thesis investigates the idea of creating a genetic algorithm able to perform clustering, which can be viewed as an optimization problem.
The aim of this thesis was to make the genetic clustering algorithm as general and automatic as possible. The end result is an algorithm that takes as input a data set and an approximate value of the number of clusters wanted, and as output gives a clustering. The algorithm is able to handle categorical values, missing samples and is also able to weight the variables in order to decide which ones that should have the greatest impact on the clustering.
The algorithm is tested on 2- and 4-dimensional data sets with clusters of different shapes and sizes. All tests made, with focus on one or many of the algorithms different features, gave results showing reasonable clustering solutions.
The use of a genetic algorithm to perform clustering seems very promising due to the algorithms ability to nd quality clusters fullling multiple objectives.}},
author = {{Karlsson, Louise}},
language = {{eng}},
note = {{Student Paper}},
title = {{Multi-objective Optimization for Clustering using Evolutionary Computing}},
year = {{2014}},
}