RGCNN: Precise Fake News Detection on Social Media Using Random Group Coupled Neural Networks

The explosion of unverified information on social media platforms has made it more important than ever to identify and anticipate false news, which poses serious hazards to democratic processes and society as a whole. The challenge is exacerbated by the lack of standardized benchmark datasets, which impedes the development of effective countermeasures against misinformation. This research introduces a novel approach for fake news classification and prediction through the Random-Coupled Neural Network with a Group Teaching Algorithm (RGCNN). Data is sourced from the LIAR and ISOT datasets and undergoes comprehensive preprocessing, including tokenization, data cleaning, removal of stopwords, stemming, and lemmatization. Feature extraction is performed using a Split Multi-Axis Vision Transformer. The Random-Coupled Neural Network (RCNN) is utilized for precise fake news detection, with its performance further refined by applying Group Teaching Optimization (GTO). 99.6% on the LIAR dataset and 99.5% on the ISOT dataset show how accurate the suggested method is.