WSEAS Transactions on Information Science and Applications
Print ISSN: 1790-0832, E-ISSN: 2224-3402
Volume 9, 2012
Isotonic Separation with an Instance Selection Algorithm Using Softset: Theory and Experiments
Authors: , ,
Abstract: In supervised machine learning, a training set containing labeled instances is taken by a learning algorithm to construct a model that is subsequently used for determining the class label of new instances. Isotonic separation is a supervised machine learning technique in which classification is represented as a Linear Programming Problem (LPP) with an objective of minimizing the number of misclassifications. It is computationally expensive to solve the LPP using traditional methods for a large dataset. Characteristics of the training set such as size, presence of noisy data, influence the learning algorithm and classification performance. To resolve this issue, this paper introduces a new linearithmic time algorithm called Soft set based instance selection algorithm (SOFIA) which provides a condensed dataset for a learning algorithm. And, a hybrid classification algorithm, SOFIA-IS which utilizes SOFIA for instance selection and isotonic separation (IS) for classification is introduced. Two sets of experimental studies are conducted on Wisconsin Breast Cancer dataset and the results are reported. First, experiments are performed on SOFIA-IS and the results are compared with isotonic separation and its variants. Then experiments are done on state of the art machine learning techniques by including SOFIA for instance selection and the results are compared with same techniques without SOFIA. Experimental and statistical results show that the condensed sets obtained by SOFIA are optimum, and SOFIA-IS and SOFIA based machine learning techniques are better in terms of classification accuracy, time and space complexity.