GAN Based Unsupervised Learning for Rapid Fake News Detection

Rapid Increase of fake news across social media platforms has come as a significant problem to various societies around the world. This research aims to conduct a detailed examination on Generative Adversarial Networks as applied to fake news detection, including its effectiveness and advantages over other machine learning approaches. We deploy a novel algorithm using GAN and calculate bilingual evaluation understudy (BLEU) score. The BLEU score analysis of proposed methodology peaks at 0.58 which show significant improvement in efficacy of fake news detection using text-based methodology.