Predicting the Performance in Learning and Recommendations to Improve Hearing Impaired Students in Special Education

Abstract

Education was one of the fundamental need and rights for all people across the world. Every government formulates different schemes to ensure education for all as it results in the countries growth on various aspects. The people who are physically impaired (PI) are also included in these aspects. The performance of those students requires continuous monitoring to acknowledge their attention towards studies and to guide them towards better academic achievements. In this paper, the Recurrent Neural network (RNN) and Hybrid firefly – particle (HFP) algorithm based novel predictor is proposed to predict semester performance of the hearing-impaired students. The RNN algorithm predict the performance of the student and HFP is involved to optimize the prediction performance that may suffer from convergence error. The proposed model was evaluated for its accuracy at both the testing and training phase. The model was initially trained with 80% of data and tested with 20% of it. The proposed model was evaluated for its accuracy at both the testing and training phase. The outcome showed that the MSE loss in training is 0.05 with testing RMSE value of 0.24. The proposed model can be enhanced to predict the drop out probability for the PI students in future.