Student Academic Performance Evaluation Using Efficient Spiking Neural Network

Abstract

Education data mining allows educational organizations to operate efficiently and effectively by using information related to all its stakeholders. The study assists to build recommendation engine and alert the students in different stages. In the study, are developing a Efficient Spiking Neural Network for determining student performance. Initially, the database is obtained from the open-source system. The proposed approach is acting in systemic steps for determining the academic performance of the students while considering both classification and clustering students' data. The data is gathered based on their behaviours, academic features and demographics characteristics. Once the dataset is collected, so that initial pre-processing technique is performed, this involves the data cleansing, data reduction, data transformation and feature selection. The cleaning data is sent to the clustering method for unifying data. In the pre-processed data, game based k-means is applied. Lastly, the proposed classifier ESNN is applied for classifying the student performance. The proposed approach consists of Spiking Neural Network and Ebola Optimization Algorithm. The proposed approach is implemented in MATLAB and evaluation is done in terms of performance matrices such as accuracy, precision, recall, sensitivity, and F_Measure. The proposed approach is compared to conventional approaches such as Deep Neural Network, Adaptive Neuro Fuzzy Interference System, and Artificial Neural Network.