Data Mining and Knowledge Engineering

Avani J. Raval ¹, Amit P. Ganatra ¹, C. K. Bhensdadia ², Yogeshwar P. Kosta ³

Affiliations
1 Charotar Institute of Technology Changa, Gujarat, IN
2 Dharmsinh Desai University, Nadiad, Gujarat, IN
3 Charotar Institute of Technology Changa, Gujarat, IN

Abstract

A common way to address a multi-class classification problem is to design a model that consists of hand picked binary classifiers and to combine them so as to solve the problem. Error-Correcting Output Codes (ECOC) is one such framework that deals with multi-class classification problems. Recent works in the ECOC domain has shown promising results demonstrating improved performance. Therefore, ECOC framework is a powerful tool to deal with multi-class classification problems. The error correcting ability improve and enhance the generalization ability of the base classifiers. This paper introduces state-of-the-art coding (one-versus-one, one-versus-all, dense random, sparse random, DECOC, forest-ECOC, and ECOC-ONE) and decoding designs (hamming, Euclidean, inverse hamming, laplacian, β-density, attenuated, loss-based, probabilistic kernel-based, and loss weighted) perspectives along with empirical study of ECOC following comparison of various ECOC methods in the above context. Towards the end, our paper consolidates details relating to comparison of various classification methods with Error Correcting Output Code method available in weka, after carrying out experiments with weka tool as a final supplement to our studies.

Keywords

Coding, Decoding, Error Correcting Output Codes, Multi-class Classification.

Full Text

     

Chintan M. Bhatt ¹, Amit P. Ganatra ¹, C. K. Bhensdadia ², Yogeshwar P. Kosta ¹

Affiliations
1 Charotar Institute of Technology Changa, Gujarat, IN
2 Dharmsinh Desai University, Nadiad, Gujarat, IN

Abstract

Intrusion is defined as an invasion that consists of set-of-actions that compromise upon the integrity, confidentiality or availability of data-resource/s. Therefore, intrusion detection is an important task when dealing with an information infrastructure for security. A major challenge in intrusion detection is to unearth intrusions that happen almost instantaneously and thereafter lay embedded, to be discovered, in vast scattered resources in a normally operating real-time communication environment. Data mining process working on intrusion detection is to identify valid, novel, potentially useful, and ultimately understandable patterns in massive data. Thus, it can be understood that, it is challenging as well as demanding to apply data mining techniques to detect intrusions of various types in an information infrastructure resource/s. To start with, our paper discusses different intrusion detection techniques that brings out and presents the underlying concepts and associated application of data mining approaches as an applied tool against intrusion detection system. Techniques include, Support Vector Machines (SVMs) that was designed and utilized as classifiers for binary classification/s, and helped to solve multi-class problems. In this paper we bring in the fusion of Decision-Tree and Support Vector Machine (DT-SVM) which combines and reinforce in an effective way for solving multi-class problems in the information resource domain. This method has the potential, as confirmed in our findings, to decrease the training and testing time, contributing to increased efficiency of the system. The construction order of binary tree significantly influences classification performance. Towards the end of the paper we report aspects relating to development of an algorithm that combines to produce a Tree structured multi-class SVM as an intrusion detection data mining technique, which has been applied successfully for the purpose of classifying data that aid the process of intrusion detection.

Keywords

Ant-Miner, COD (Common Outlier Detection), Decision Tree, Fuzzy C-Means, K-Means, MACO, Support Vector Machine (SVM) and Decision-Tree and Support Vector Machine (DT-SVM).

Full Text

     

Ankit P. Vaishnav ¹, Amit P. Ganatra ², C. K. Bhensdadia ¹

Affiliations
1 Department of Computer Engineering, Dharmsinh Desai University, Nadiad, Gujarat, IN
2 Charotar University of Science Technology (CHARUSAT), Education Campus, Changa, Gujarat, IN

Abstract

Support Vector Machines (SVMs) have been extensively researched in the data mining and machine learning communities for the last decade and actively applied to applications in various domains. SVMs are typically used for learning classification and regression tasks. Two special properties of SVMs are that they achieve (1) high generalization by maximizing the margin and (2) support an efficient learning of nonlinear functions by kernel trick. Many algorithms and their improvements have been proposed to train SVMs. This paper presents a comprehensive description of various SVM methods and compares SVM classifier with other classification methods.

Keywords

Classifiers, Machine Learning, Predictive Accuracy, Support Vector Machine (SVM).

Full Text