Informatics Publishing Limited

K. Poornamala ¹, R. Lawrance ²

Affiliations
1 Computer Science Department, Ayya Nadar Jankai Ammal College, Sivakasi, IN
2 Computer Application Department, Ayya Nadar Janaki Ammal College, Sivakasi, IN

Abstract

In recent years, data mining is an important aspect for generating association rules among the large number of itemsets. Association rule mining is one of the techniques in data mining that has two subprocess. First, the process called as finding frequent itemsets and second process is association rules mining. In this subprocess, the rules with the use of frequent itemsets have been extracted. Researchers developed a lot of algorithms for finding frequent itemsets and association rules. The frequent pattern technique only used for very large dataset and it takes large memory space tree creation. The major advantage of using Genetic Algorithm is that it perform global search and the time complexity is less compared to other algorithms. In this paper, first, GA is used to optimize the large dataset. Second, the improved frequent pattern tree is used to mine the frequent itemset without generating conditional FP-tree.

Keywords

Association Rule Mining, Data Mining, Frequent Itemset Mining, FP-Tree, Genetic Algorithm.

Full Text

     

S. Sumathi ¹, R. Lawrance ²

Affiliations
1 Department of Computer Science, Ayya Nadar Janaki Ammal College, Sivakasi, IN
2 Ayya Nadar Janaki Ammal College, Sivakasi, IN

Abstract

In the earlier days, the association rule mining is used for Market Basket analysis to find the regularity in purchasing behavior of customer. Association Rule Mining (ARM) is one of the functionalities in Data Mining, to find the relationships among the items in a particular set of itemsets. There are huge numbers of algorithms to find relationships among the items. In this paper we introduce a new Boolean algebraic algorithm for finding frequent itemsets and deriving the association rules in a large transaction database. It has two phases. In the first phase, it finds the frequent itemsets. In the second phase, by using the Boolean AND and XOR operator, it derives the association rules from the founded frequent itemset in first phase. This algorithm mines the association rules efficiently than Apriori.

Keywords

Association Rule Mining, Boolean Algebra, Data Mining, Frequent Item Set Mining.

Full Text

     

K. Suriya Prabha ¹, R. Lawrance ²

Affiliations
1 Department of Computer Science, Ayya Nadar Janaki Ammal College, Sivakasi-626124, Tamil Nadu, IN
2 Department of MCA, Ayya Nadar Janaki Ammal College, Sivakasi-626124, Tamil Nadu, IN

Abstract

The problem of mining quantitative data from large transaction database is considered to be an important critical task. Researchers have proposed efficient algorithms for mining of frequent itemsets based on Frequent Pattern (FP) tree like structure which outperforms Apriori like algorithms by its compact structure and less generation of candidate itemsets mostly for binary data items from huge transaction database. Fuzzy logic softens the effect of sharp boundary intervals and solves the problem of uncertainty present in data relationships. This proposed approach integrates the fuzzy logic in the newly invented tree-based algorithm by constructing a compact sub-tree for a fuzzy frequent item significantly efficient than other algorithms in terms of execution times, memory usages and reducing the search space resulting in the discovery of fuzzy frequent itemsets.

Keywords

Association Rule Mining, Data Mining, Fuzzy Frequent Itemset, Fuzzy Logic, Membership Function.

Full Text

     

R. Syed Ali Fathima ¹, R. Lawrance ²

Affiliations
1 Department of Computer Science, Ayya Nadar Janaki Ammal College, Sivakasi, IN
2 Ayya Nadar Janaki Ammal College, Sivakasi, IN

Abstract

In recent years, most of the researchers discover the association rules among itemsets for large database become popular. It is one of the techniques used to mine the database. There are several efficient algorithms are produced different search strategies for finding the frequent itemsets and those algorithms are very popular in the association rule mining. Many association rule mining algorithms suffer from many problems when mining the massive datasets. Some of the major problems are: (1) repetitive scans (2) huge computation time takes during the candidacy generation and (3) high memory space required. This paper, proposed a hybrid search algorithm are developed for mining multilevel association rules and it improve the performance of algorithm. This algorithm is named as Partition-COFI Tree i.e., PC tree. The proposed algorithm works faster compared to other algorithm. It improves the performance of search space, I/O and CPU time.

Keywords

Association Rule Mining, Cofi Tree, Data Mining, Partition Algorithm.

Full Text

     

C. Anand ¹, R. Lawrance ²

Affiliations
1 Department of Computer Applications, Ayya Nadar Janaki Ammal College, Sivakasi-626 124, Tamil Nadu, IN
2 Department of Computer Applications, Ayya Nadar Janaki Ammal College, Sivakasi-626 124, Tamil Nadu, IN

Abstract

Face Recognition is a significant research area since it has plenty of application domains in pattern recognition, image processing, biometrics etc. Researchers contributed lot of algorithms and techniques to uncover the mask of face recognition arena. In our research, a Hidden Markov Model (HMM) based face recognition system along with Singular Value Decomposition (SVD) coefficients is discussed. Human face is divided into seven facial regions and a few quantized SVD Coefficients were trained to choose the facial features. Median Filtering is used as a preprocessing operation for efficient computation. Observation vectors are generated by dividing each face image into overlapping blocks and SVD coefficients acts as a base for constructing the observation sequence. Due to the discrete nature of HMM, quantization process is introduced to model the continuous observation vectors. The system is trained and tested on ORL face database consist of 400 images of 40 persons in Portable Gray Map (.pgm) format. Five face images of a person are considered for training the system and tested against 200 unseen faces. Our system achieves a recognition rate of 99.5% with a computational speed of 0.22 seconds per image. Choosing SVD coefficients as features increases the efficiency, in turn reducing the complexity. Experimental results revealed that our proposed system outperforms many of the traditional HMM based face recognition methods.

Keywords

Face Recognition, Hidden Markov Models (HMM), Median Filtering, Pattern Recognition, Singular Value Decomposition (SVD) Coefficients.

Full Text

     

K. Aruna Devi ¹, T. Marimuthu ¹, R. Lawrance ²

Affiliations
1 Computer Application, ANJA College, Sivakasi, Tamilnadu, IN
2 ANJA College, Sivakasi 626 124, Tamilnadu, IN

Abstract

The core objective of this paper is to analyze the Dengue Serotypes using online biological databases & tools and to generate a novel gracious tool to carry out sequence analysis using bioinformatics algorithms and concepts on Dengue virus sequences or other organisms that will be helpful in attaining knowledge for new invention. The Dengue virus is a member of the family Flavi virus. It is transmitted to people through the bite of the mosquitoes “Aedes aegypti" and “Aedes albopictus”. Dengue virus is now believed to be the most common anthropod-borne disease (an infectious disease carried by insect vectors) in the world and the Dengue fever is also called as “Break Bone Fever”. Dengue is mainly found in the tropics because the mosquitoes require a warm climate. A major fear of epidemiologists is that the mosquitoes will develop resistance to cooler climates and then be able to infect people living in any climate. Predicting the relationship between Dengue Serotypes will definitely help the Biotechnologists and Bioinformaticians to move one step forward in discovering vaccine for Dengue. For that the tool “Sequence Miner” created with computational intelligence is much helpful. Computational intelligence comprises practical adaptation concepts, paradigms, algorithms and implementations that enable or facilitate appropriate actions (intelligent behavior) in complex and changing environment.

Keywords

Dengue Serotype, Fasta, Phylogenetic Tree, Sequence Alignment.

Full Text

     

C. Anand ¹, R. Lawrance ¹

Affiliations
1 PG Department of Computer Applications, Ayya Nadar Janaki Ammal College, Sivakasi-626 124, Tamil Nadu, IN

Abstract

Face Recognition stands high as a significant research area since it has plenty of application domains in pattern recognition, image processing, biometrics etc. Researchers contributed lot of algorithms and techniques to uncover the mask of face recognition arena. In this paper, a Left-Right Hidden Markov Models (HMM) based face recognition algorithm along with Singular Value Decomposition (SVD) Coefficients is discussed. Human face is divided into seven facial regions and a small number of quantized SVD Coefficients were trained to choose the facial features. Order Statistic Filtering is used as a preprocessing operation for efficient computation. Using SVD Coefficients, a face is considered as a numerical sequence representing block of images which can be easily modeled by discrete HMM. The system is tested on Olivetti Research Laboratory (ORL) face database consist of 400 images of 40 persons in .pgm format. For training, five face images of a person are considered and our proposed system achieves a recognition rate of 96.5% with a computational speed of 0.22 seconds per image. The experimental results reveal that our proposed system outperforms many of the traditional face recognition methods tested on ORL database.

Keywords

Face Recognition, Hidden Markov Models (HMM), Order Statistic Filtering, Pattern Recognition, Singular Value Decomposition (SVD) Coefficients.

Full Text

     

M. Lydia Packiam Mettilda ¹, R. Lawrance ², R. Kalaivani ³

Affiliations
1 Department of Computer Application, Ayya Nadar Janaki Ammal College, Sivakasi, IN
2 Computer Application Department, Ayya Nadar Janaki Ammal College, Sivakasi, IN
3 Department of Computer Science and Information Technology, Ayya Nadar Janaki Ammal College, Sivakasi, IN

Abstract

With the advancement in technology, one thing that concerns the world and especially in the developing countries is the tremendous increase in population. With such a rapid rate of increase, it is becoming difficult to recognize each and every person because we have to keep up photos either in digital or hard copy format of every person at different time periods of his/her life. Sometimes database has the required information of that particular person, but it’s of no use as it is now obsolete. Deciding age of a person from digital photography is an intriguing problem. Age changes cause many variations in visible of human faces. Many aspects affect the appearance of a person’s face during the process of growing older. The aging process will explain with many factors such as health, living style, living place and weather condition etc. Face is a non-intrusive recognition, without user co-ordination able to recognize the person. Age classification system is generally composed of feature extraction and classification. It is used to estimate the age of a person from his/her face features. For the aging feature extraction, face images interpreted as decomposition of optimized wavelet transform with 49 feature vectors using Daubechies wavelet and the classifier of supervised neural network to discriminate the ranges of ages. The work is to classify the age range into child (1-10), teenage (11-20) young (21-30), middle aged (31-50) and old (51 and above).

Keywords

ASM, Wavelet, Age Classification, Neural Network.

Full Text

     

R. Kalaivani ¹, C. Devi Arockia Vanitha ², R. Lawrance ³, M. Lydia Packiam Mettilda ⁴

Affiliations
1 Department of Computer Science & Information Technology, Ayya Nadar Janaki Ammal College, Sivakasi 626 124, Tamilnadu, IN
2 Computer Science Department, The S.F.R. College for Women, Sivakasi 626123, Tamilnadu, IN
3 MCA Department, Ayya Nadar Janaki Ammal College, Sivakasi 626 124, Tamilnadu, IN
4 Department of Computer Application, ANJA College, Sivakasi 626 124, Tamilnadu, IN

Abstract

Most of the real world problems in engineering, medicine, industry, science and business also involve data classification. Classification is a supervised machine learning technique used to predict group membership for data instances. Pattern classification problems belong to the category of supervised learning. Pattern Classification involves assigning a label to a given input data. Neural Networks are an effective tool in the field of pattern classification, using training and testing data to build a model. Training neural networks in classification problems, especially when biological data are is a very challenging task. The protein superfamily classification problem, which consists of determining the superfamily membership of a given unknown protein sequence, is very important for a biologist for many practical reasons, such as drug discovery, prediction of molecular function and medical diagnosis. The objective of this work is creating a classification model for classifying data using Multilayer feed forward network. It contain two phases. First, classifier model was build for iris plant classification. Second, classifier model was build for protein sequence classification to know the organism of protein and family of the given protein sequence.

Keywords

Data Mining, Pattern Classification, Neural Network, Back Propagation, Iris Plant, Bioinformatics, Protein Sequence.

Full Text

     

K. Poornamala ¹, R. Lawrance ²

Affiliations
1 Department of Computer Science, Ayya Nadar Janaki Ammal College, IN
2 Department of Computer Applications, Ayya Nadar Janaki Ammal College, IN

Abstract

In recent years, data mining is an important aspect for generating association rules among the large number of itemsets. Association rule mining is one of the techniques in data mining that that has two sub processes. First, the process called as finding frequent itemsets and second process is association rules mining. In this sub process, the rules with the use of frequent itemsets have been extracted. Researchers developed a lot of algorithms for finding frequent itemsets and association rules. Recently association rule mining systems have been designed using a combination of soft computing techniques.

Keywords

Data Mining, Association Rule Mining, Frequent Itemset Mining, Genetic Algorithm.

Full Text

     

V. Shanmugarajeshwari ¹, R. Lawrance ¹

Affiliations
1 Department of Computer Applications, Ayya Nadar Janaki Ammal College, Sivakasi, IN

Abstract

Data mining is one of the potential research fields regarding interdisciplinary aspects. Educational data mining is developing discipline in the present scenario. Classification techniques in the data mining plays an important role in the area of educational data mining. The main goal regarding this work is to predict the teachers’ performance by using the relevant features. The proposed methodology consists of the phases like preprocessing, attribute selection, classification based on decision tree and performance evaluation. In the data preprocessing phase, the missing values have been filluped. The attributes are converted into a categorized format using the categorization MapReduce process. The gain ratio with MapReduce is the best method for feature selection, since it selects technique extracted the relevant attributes in an accurate manner, which takes less time compares to the other feature selection methods. This paper presents a MapReduce algorithm on the classification structure by using C5.0 algorithm, aiming to accumulate time and obtain high accuracy on huge students’ and teachers’ datasets. The classification process based on C5.0 MapReduce algorithm is resulted with good classification accuracy.

Keywords

Educational Data Mining, Students’ and Teachers’ Dataset, MapReduce, Classification, C5.0 Algorithm, Big Data Mining.

Full Text