ICTACT Journal on Soft Computing

N. Sharmila Banu ¹, S. Suganya ¹

Affiliations
1 Department of Computer Science, Rathnavel Subramaniam College of Arts and Science, IN

Abstract

Epilepsy is the neural disorder that occurs in the individual mind which affects nearly 50 million people around the world. It is also said to be the universal disorder which affects all ages. The disturbance that occurs in the nervous system causes seizure. The classification of epileptiform activity in the EEG plays an essential role in the identification of epilepsy. To extract the relevant information and to improve the accuracy level from the given EEG signals, Fuzzy Based Cuckoo Search (FCS) and ant colony optimization (ACO) methods are planned to select the related and best information’s. Finally utilizes the Ensemble Hybrid Classifiers (EHCs) which combine the procedure of Modified Convolutional Neural Network (MCNN), Improved Relevance Vector Machine (IRVM) and Logistic Regression (LR) classifiers for analysis of EEG signals. The planned effort is implemented to notice the irregularity in three different levels of EEG signals (normal, affected and unaffected).

Keywords

Epilepsy, Seizure, Ant Colony Optimization, Convolution Neural Network.

Full Text

     

K. S. Narayanan ¹, S. Suganya ²

Affiliations
1 Department of MCA, Rathnavel Subramaniam College of Arts and Science, IN
2 Department of Computer Science, Rathnavel Subramaniam College of Arts and Science, IN

Abstract

The employment of internet and social media has remodeled behavioral aspects of consumer or student communities and methods in which organizations or educational institutions perform their business pattern. Both social and digital marketing put forwards efficient scopes to educational institutions by way of reduced costs, enhanced brand perception and elevated sales. Nevertheless, notable disputes prevail from obstructive electronic word-of-mouth and invasive and annoying online brand existence. Nowadays, students use online promotions to know about best universities for education globally. This university choice and students’ feedback observed by student-experience shared across social media platforms. Several methods have been employed for selecting the university but not providing accurate information. This paper is motivated towards applying Machine Learning for learning, analyzing and classifying the student information based on the student experience by means of tweets in twitter. The twitter data with student tweets is collected from benchmark twitter dataset and applied to the proposed method, Class-wise Linear Discriminant and Regression-based Binarized Nearest Learning (CLD-RBNL). The CLD-RBNL method is split into two sections. First, preprocessing and relevant feature selection (i.e. tweets) are acquired by employing Class-wise Linear Discriminant-based Feature Selection (CLDFS) model to obtain dimensionality reduced tweets. To this result, Regression-based Binarized Nearest Neighbor model is applied for maximum lead generation. The CLD-RBNL method is compared with other state-of-the-art methods and found to outperform in terms of sensitivity, specificity, processing time, lead generation accuracy and error rate.

Keywords

Class-Wise, Linear Discriminant, Feature Selection, Digital Marking, Educational Services, Regression, Binarized Nearest Neighbor.

Full Text