Optimization of Association Rule Mining for Mammogram Classification
P. Niraj, and U. Sonar. International Journal of Image Processing (IJIP), 11 (3):
64-87(June 2017)
Abstract
Authors presents concept of image mining, an extension of data mining, for discovering image data relationship from a large collection of mammograms images. Association rule mining is the process of discovering useful and interesting rules from large datasets based on user specified minimum support and confidence values. The set of all possible item sets grows exponentially with the number of items in the database. These constraints lead to exponential search space and dataset dependent minimum support and confidence values. It generates a huge number of unnecessary rules from frequent item sets and results in weak mining performance. The authors propose two association rule optimization techniques for overcoming these problems. The first graph theory approach (OARGT) is based on objective function such that graph generated by the optimized rules is a simple graph with simple walk. The second approach is based on Multi-Objective Genetic algorithm (MOGA) with adaptive crossover and mutations (MOGAACM). Traditionally, input to ARM classifier is in binary format. The proposed system uses variable feature quantization and relevant feature selection. In MOAGAACM, ranks are assigned to rules as per fitness function. The rules with highest rank, low crossover and mutation rates are assigned and vice versa. Experimental results show that, MOGAACM generates more effective and strong association rules compared with objective function using graph theory and achieves 89.08% and 43.60 % reduction in association rules for benign and malignant class respectively for MIAS database and 80.13 % and 79.60 % reduction in association rules for benign and malignant class respectively for DDSM medical image database respectively. The Graph theory achieves 9 % reduction in rules for MIAS data base. Authors propose class identification using Strength of Classification Algorithm (CISCA) for classification of mammogram image into benign and malignant classes. The classification accuracy measures reported are 91.66 % for MIAS database using graph theory and 95.45 % and 92.5 % for MIAS and DDSM respectively using MOGAACM.
%0 Journal Article
%1 nira2017optimization
%A Niraj, Poonam
%A Sonar, Udhav Bhosle
%D 2017
%J International Journal of Image Processing (IJIP)
%K Adaptive Algorithm Association Confidence, Crossover, Fitness Function, Genetic Graph Mining, Multi Mutation, Objective Rule Support, Theory,
%N 3
%P 64-87
%T Optimization of Association Rule Mining for Mammogram Classification
%U http://www.cscjournals.org/library/manuscriptinfo.php?mc=IJIP-1122
%V 11
%X Authors presents concept of image mining, an extension of data mining, for discovering image data relationship from a large collection of mammograms images. Association rule mining is the process of discovering useful and interesting rules from large datasets based on user specified minimum support and confidence values. The set of all possible item sets grows exponentially with the number of items in the database. These constraints lead to exponential search space and dataset dependent minimum support and confidence values. It generates a huge number of unnecessary rules from frequent item sets and results in weak mining performance. The authors propose two association rule optimization techniques for overcoming these problems. The first graph theory approach (OARGT) is based on objective function such that graph generated by the optimized rules is a simple graph with simple walk. The second approach is based on Multi-Objective Genetic algorithm (MOGA) with adaptive crossover and mutations (MOGAACM). Traditionally, input to ARM classifier is in binary format. The proposed system uses variable feature quantization and relevant feature selection. In MOAGAACM, ranks are assigned to rules as per fitness function. The rules with highest rank, low crossover and mutation rates are assigned and vice versa. Experimental results show that, MOGAACM generates more effective and strong association rules compared with objective function using graph theory and achieves 89.08% and 43.60 % reduction in association rules for benign and malignant class respectively for MIAS database and 80.13 % and 79.60 % reduction in association rules for benign and malignant class respectively for DDSM medical image database respectively. The Graph theory achieves 9 % reduction in rules for MIAS data base. Authors propose class identification using Strength of Classification Algorithm (CISCA) for classification of mammogram image into benign and malignant classes. The classification accuracy measures reported are 91.66 % for MIAS database using graph theory and 95.45 % and 92.5 % for MIAS and DDSM respectively using MOGAACM.
@article{nira2017optimization,
abstract = {Authors presents concept of image mining, an extension of data mining, for discovering image data relationship from a large collection of mammograms images. Association rule mining is the process of discovering useful and interesting rules from large datasets based on user specified minimum support and confidence values. The set of all possible item sets grows exponentially with the number of items in the database. These constraints lead to exponential search space and dataset dependent minimum support and confidence values. It generates a huge number of unnecessary rules from frequent item sets and results in weak mining performance. The authors propose two association rule optimization techniques for overcoming these problems. The first graph theory approach (OARGT) is based on objective function such that graph generated by the optimized rules is a simple graph with simple walk. The second approach is based on Multi-Objective Genetic algorithm (MOGA) with adaptive crossover and mutations (MOGAACM). Traditionally, input to ARM classifier is in binary format. The proposed system uses variable feature quantization and relevant feature selection. In MOAGAACM, ranks are assigned to rules as per fitness function. The rules with highest rank, low crossover and mutation rates are assigned and vice versa. Experimental results show that, MOGAACM generates more effective and strong association rules compared with objective function using graph theory and achieves 89.08% and 43.60 % reduction in association rules for benign and malignant class respectively for MIAS database and 80.13 % and 79.60 % reduction in association rules for benign and malignant class respectively for DDSM medical image database respectively. The Graph theory achieves 9 % reduction in rules for MIAS data base. Authors propose class identification using Strength of Classification Algorithm (CISCA) for classification of mammogram image into benign and malignant classes. The classification accuracy measures reported are 91.66 % for MIAS database using graph theory and 95.45 % and 92.5 % for MIAS and DDSM respectively using MOGAACM.},
added-at = {2018-12-14T11:14:34.000+0100},
author = {Niraj, Poonam and Sonar, Udhav Bhosle},
biburl = {https://www.bibsonomy.org/bibtex/2f2863796e0de5350e10b55e2cbfc8448/cscjournals},
interhash = {a9022ae0f60d4db7067ef8f828a6ec3b},
intrahash = {f2863796e0de5350e10b55e2cbfc8448},
issn = {1985-2304},
journal = {International Journal of Image Processing (IJIP)},
keywords = {Adaptive Algorithm Association Confidence, Crossover, Fitness Function, Genetic Graph Mining, Multi Mutation, Objective Rule Support, Theory,},
language = {English},
month = {June},
number = 3,
pages = {64-87},
timestamp = {2018-12-14T11:14:34.000+0100},
title = {Optimization of Association Rule Mining for Mammogram Classification},
url = {http://www.cscjournals.org/library/manuscriptinfo.php?mc=IJIP-1122},
volume = 11,
year = 2017
}