Fake News Detection Using Evolutionary Ensemble Deep Learning

In today's era, the digital consumer base is increasing exponentially around the globe every year. This has created a surge of information containing fake and real information throughout the network and enabled easy and seamless dissemination. The fake news creates an illusory scene aimed at the personal, political, social, or monetary gains of a person or organization. Thus, detecting false or misleading information from the plethora of information is a substantial task. Using the Evolutionary Algorithm and an ensemble deep learning model over the dataset Gossipcop, we have used a hybrid model for the classification of real and fake news. Four popular deep learning models, namely LSTM, BiLSTM, CNNLSTM, and CNNBiLSTM, have been ensembled, powered by a novel version of an Evolutionary Genetic Algorithm called Batch Genetic Algorithm. The proposed model provides an overwhelming accuracy of 0.8599. Also, to substantiate the proposed model, it has been compared with twelve models based on different a technique which affirms the supremacy of the proposed model.