Leveraging the Robust Capability of Modified Multi-Layer Gated Recurrent Units for Fake News Detection

Fake news transmission poses a major hazard to public debate and information integrity. The rise of fake news on many platforms has emphasized the vital demand for verifying the accuracy of information. In recent years, the application of neural network designs, particularly recurrent models, has shown promise in determining the authenticity of news reports and content. Gated Recurrent Units (GRUs), a variant of recurrent neural networks, stand up as a potential solution due to their inherent benefits in processing sequential input and finding specific patterns within textual content. Their potential to maintain crucial information while selectively decreasing noise enables more powerful identification of false patterns concealed inside news articles. Furthermore, GRUs offer computational efficiency compared to more advanced recurrent architectures, enabling faster training and inference. This research seeks to examine and illustrate the utility of gated recurrent units in the realm of fake news identification. The proposed GRUs-based modified multi-layer deep architecture model correctly predicts the news with an accuracy of 99.59%, 99.59% precision, and a recall of 99.59%. The proposed work surpasses nearly all recent comparable research in terms of accuracy and efficiency.