Deep Fake Video Detection Using Transfer Learning Approach

Area of Science:

• Computer Science
• Artificial Intelligence
• Cybersecurity

Background:

• The rapid spread of fake news and sophisticated deep fakes online poses significant societal and political threats.
• The accessibility of deep fake generation tools necessitates advanced computational methods for detection.
• Increased complexity in tampering techniques makes detecting fake content challenging.

Purpose of the Study:

• To propose a novel framework for detecting fake videos.
• To evaluate the generalizability of the proposed model using unseen test data.
• To analyze the impact of residual image input on detection accuracy.

Main Methods:

• Utilizing transfer learning in autoencoders.
• Implementing a hybrid model combining Convolutional Neural Networks (CNN) and Recurrent Neural Networks (RNN).
• Analyzing the effect of residual image input on model performance.

Main Results:

• The proposed framework demonstrates effectiveness in detecting fake videos.
• Transfer learning significantly enhances the accuracy of the detection model.
• The generalizability of the model was confirmed through testing on unseen data.

Conclusions:

• The developed framework offers a robust solution for identifying sophisticated fake videos.
• Transfer learning is a crucial component for improving the performance of deep fake detection systems.
• The study highlights the importance of advanced computational approaches to combat misinformation.