A Lightweight Algorithm for Detecting Fake Multimedia Contents on Social Media

Abstract

The significant growth of the fourth industrial revolution (Industry 4.0) coupled with the widespread adoption of social media across the world has initiated new challenges that deserve the attention of researchers and industry leaders especially in detecting and preventing fake multimedia contents on social media. The forging of multimedia contents like videos and images for malicious activities is gradually becoming very rampant and this has serious psychological, health, political and economic consequences on the targeted individuals or close associates of the victims. The application of deepfake algorithms to make manipulated videos and images has contributed in making it very difficult to identify fake videos and images from the real multimedia contents. The availability of the internet and social media has made the spread of deepfake videos and images very fast and at an alarming rate. This research work understanding the dire need to detect deepfake videos and images (multimedia contents) proposes a lightweight algorithm to detect deepfake videos and images on social media platforms. The need for a lightweight algorithm is essential to enable low computational devices to be able to apply the algorithm without computational challenges and overheads. The proposed model has demonstrated a significant reduction in the computational and time complexities. The research work also presented a comparative analysis of some selected deep learning models with emphasis on the datasets used, their features and challenges identified.