FEATURE WEIGHTED BIPOLAR NEUTROSOPHIC CONVOLUTIONAL NEURAL NETWORK FOR PREDICTING CHILD MALNUTRITION

Abstract

The uncertainty issues have happened in prediction of malnutrition status using a deep learning model. So, solve this uncertainty issues in deep learning model by enhancing the deep learning architecture. To solve the issue of uncertainty information’s in malnutrition, Bipolar Neutrosophic Convolutional Neural Networks (BNCNN) is developed for extracting different deep features to generate predictive uncertainty estimates. In this bipolar neutrosophic set characterized by true positive, true negative, false positive and false negative. This model used all set of features from the given dataset. So unimportant features also considered to predict the accuracy result. Whenever involved unimportant data, the model produced the result with less accuracy. In this proposed work using Regularization techniques to remove the less important features using weighted softmax algorithm and considered only subset of features from the given dataset and produce the more accuracy. Compared to Bipolar Neutrosophic Convolutional Neural Networks, the proposed model of the Feature weighted Bipolar neutrosophic model is produced more accuracy results.