ICTACT Journal on Soft Computing

V. Shyamala Susan ¹, T. Christopher ²

Affiliations
1 Department of Computer Science, Government Arts College, Udumalpet, IN
2 Department of Computer Science, Government Arts College, Coimbatore, IN

Abstract

Privacy preservation is that the most targeted issue in information publication, because the sensitive data shouldn't be leaked. For this sake, several privacy preservation data mining algorithms are proposed. In this work, feature selection using evolutionary algorithm and data masking coupled with slicing is treated as a multiple objective optimisation to preserve privacy. To start with, Genetic Algorithm (GA) is carried out over the datasets to perceive the sensitive attributes and prioritise the attributes for treatment as per their determined sensitive level. In the next phase, to distort the data, noise is added to the higher level sensitive value using Hybrid Data Transformation (HDT)method. In the following phase slicing algorithm groups the correlated attributes organized and by this means reduces the dimensionality by retaining the Advanced Clustering Algorithm (ACA). With the aim of getting the optimal dimensions of buckets, tuple segregating is accomplished by Metaheuristic Firefly Algorithm (MFA). The investigational consequences imply that the anticipated technique can reserve confidentiality and therefore the information utility is additionally high. Slicing algorithm allows the protection of association and usefulness in which effects in decreasing the information dimensionality and information loss. Performance analysis is created over OCC 7 and OCC 15 and our optimization method proves its effectiveness over two totally different datasets by showing 92.98% and 96.92% respectively.

Keywords

Privacy Preservation, Genetic Algorithm, Advanced Clustering Algorithm, Metaheuristic Firefly Algorithm, Hybrid Data Transformation.

Full Text

     

Saravanan Suba ¹, T. Christopher ²

Affiliations
1 Department of Computer Science, Kamarajar Government Arts College, IN
2 Department of Computer Science, Government Arts College, Coimbatore, IN

Abstract

Mining association rules in large database is one of most popular data mining techniques for business decision makers. Discovering frequent item set is the core process in association rule mining. Numerous algorithms are available in the literature to find frequent patterns. Apriori and FP-tree are the most common methods for finding frequent items. Apriori finds significant frequent items using candidate generation with more number of data base scans. FP-tree uses two database scans to find significant frequent items without using candidate generation. This proposed TR-FCTM (Transaction Reduction- Frequency Count Table Method) discovers significant frequent items by generating full candidates once to form frequency count table with one database scan. Experimental results of TR-FCTM shows that this algorithm outperforms than Apriori and FP-tree.

Keywords

Apriori, FP-Tree, TR-FCTM, Minimum Support.

Full Text

     

K. G. Nandha Kumar ¹, T. Christopher ¹

Affiliations
1 Department of Computer Science, Government Arts College, Udumalpet, IN

Abstract

An attempt has been made to evaluate the feedbacks of library users of four different libraries by using neural network based data mining techniques. This paper presents the results of a survey of users' satisfactory level on four different libraries. The survey has been conducted among the users of four libraries of educational institutions of Kovai Medical Center Research and Educational Trust. Data were collected through questionnaires. Artificial neural network based data mining techniques are proposed and applied to assess the libraries in terms of level of satisfaction of users. In order to assess the users' satisfaction level, two neural network techniques: Modified Multilayer Perceptron Network-Supervised and Modified Multilayer Perceptron Network-Unsupervised are proposed. The proposed techniques are compared with the conventional classification algorithm Multilayer Perceptron Neural Network and found better in overall performance. It is found that the quality of service provided by the libraries is highly good and users are highly satisfied with various aspects of library service. The Arts and Science College Library secured the maximum percent in terms of user satisfaction. This shows that the users' satisfaction of ASCL is better than the other libraries. This study provides an insight into the actual quality and satisfactory level of users of libraries after proper assessment. It is strongly expected that the results will help library authorities to enhance services and quality in the near future.

Keywords

Academic Libraries, Users’ Satisfaction, Data Mining, Data Classification, Artificial Neural Networks.

Full Text

     

R. Kavitha ¹, T. Christopher ¹

Affiliations
1 Department of Computer Science, PSGR Krishnammal College for Women, IN

Abstract

The electrical activity of the human heart is measured by the vital bio medical signal called ECG. This electrocardiogram is employed as a crucial source to gather the diagnostic information of a patient's cardiopathy. The monitoring function of cardiac disease is diagnosed by documenting and handling the electrocardiogram (ECG) impulses. In the recent years many research has been done and developing an enhanced method to identify the risk in the patient's body condition by processing and analysing the ECG signal. This analysis of the signal helps to find the cardiac abnormalities, arrhythmias, and many other heart problems. ECG signal is processed to detect the variability in heart rhythm; heart rate variability is calculated based on the time interval between heart beats. Heart Rate Variability HRV is measured by the variation in the beat to beat interval. The Heart rate Variability (HRV) is an essential aspect to diagnose the properties of the heart. Recent development enhances the potential with the aid of non-linear metrics in reference point with feature selection. In this paper, the fundamental elements are taken from the ECG signal for feature selection process where Bat algorithm is employed for feature selection to predict the best feature and presented to the classifier for accurate classification. The popular machine learning algorithm ELM is taken for classification, integrated with evolutionary algorithm named Self- Adaptive Differential Evolution Extreme Learning Machine SADEELM to improve the reliability of classification. It combines Effective Fuzzy Kohonen clustering network (EFKCN) to be able to increase the accuracy of the effect for HRV transmission classification. Hence, it is observed that the experiment carried out unveils that the precision is improved by the SADE-ELM method and concurrently optimizes the computation time.

Keywords

Self-adaptive Differential Evolution, Extreme Learning Machine, Bat Algorithm, Efficient Fuzzy Kohonen Clustering Network.

Full Text

     

S. Subash Chandra Bose ¹, T. Christopher ²

Affiliations
1 Department of Computer Science, Government Arts College, Udumalpet, IN
2 Department of Computer Science, Government Arts College, Coimbatore, IN

Abstract

Different clustering algorithms produce distinct sub-divisions as they apply disparate partition on the data. Hence, no single clustering algorithm is said to be optimal and therefore resulting in different partitions. To utilize the complementary nature of different partitions, ensemble clustering is used. The work in this paper focuses on producing ensembles through several clustering algorithms that perform feature extraction using deep learning and malicious tumor detection through ensemble cluster. In this study, to improve the performance and reduce the complexity involved in the malicious tumor detection process, Deep Learning Feature Extraction (DLFE) technique is presented. Furthermore, to improve the quality of results obtained, ensemble clusters namely, Normalized Spectral Cluster and Gaussian Mixture technique has been applied to the extracted features. The experimental results of the proposed technique have been evaluated and validated for performance and quality analysis on three datasets based on accuracy, sensitivity, specificity. The experimental results achieved 85.28% accuracy, 70.43% specificity, and 97.19% sensitivity, demonstrating the effectiveness of the proposed technique for identifying normal and abnormal tissues from various test images. The simulation results prove the significance in terms of quality parameters and accuracy in comparison to the state-of-the-art techniques.

Keywords

Clustering Algorithm, Deep Learning, Feature Extraction, Normalized Spectral Cluster, Gaussian Mixture.

Full Text

     

Thomas Scaria ¹, T. Christopher ²

Affiliations
1 Department of Computer Science, St. Pius X College, IN
2 Department of Information Technology, Government Arts College, Coimbatore, IN

Abstract

Data mining plays an important role in the process of classifying between the normal and the cancerous samples by utilizing microarray gene data. As this classification process is related to the human lives, greater sensitivity and specificity rates are mandatory. Taking this challenge into account, this work presents a technique to classify between the normal and cancerous samples by means of efficient feature selection and classification. The process of feature selection is achieved by Information Gain Ratio (IGR) and the selected features are forwarded to the classification process, which is achieved by ensemble classification. The classifiers being employed to attain ensemble classification are k-Nearest Neighbour (k-NN), Support Vector Machine (SVM) and Extreme Learning Machine (ELM). The performance of the proposed approach is analysed with respect to three different datasets such as Leukemia, Colon and Breast cancer in terms of accuracy, sensitivity and specificity. The experimental results prove that the proposed work shows better results, when compared to the existing techniques.

Keywords

Data Mining, Classification, Feature Selection.

Full Text