Image Fake News Prediction Based on Random Forest and Gradient-boosting Methods

Abstract

The internet technology of today makes it challenging to spread false information, particularly through photos, including fake news. In this study, fake news is identified and predicted using photos that have been altered or misrepresented. Effective detection systems are crucial because of the proliferation of false information that images might spread due to the use of image modification tools and social media. This paper provides a thorough analysis of fake news based on images. Among the main research areas are machine learning for classification models and image data embedding (feature extraction). Our novel methodology forecasts fake news in the form of altered or misleading photographs by using Random Forest and gradient-boosting algorithms to detect visual alterations such as picture editing and image synthesis. This research leverages massive image datasets from news channels and social media to train and assess predictive algorithms. Our results demonstrate that our method has strong recall and precision in identifying image-based fake news. We also discuss practical applications and real-time detection, such building tools to combat misinformation on social media and in news organizations. At 0.968 with 0.997, Gradient Boosting performs better than Random Forest.