Portfolio

Deepfakes

Just as advances in machine learning gave rise to deepfake videos, so too can modern machine learning models be put to use to reliably detect fake media content. In this project I set out to do just that – first using deep neural networks to generate deepfake videos, then using neural networks to correctly distinguish between fake and unaltered media content. In each scenario, I test the system on a range of different videos in order to evaluate the effectiveness of the proposed network.

Deepfake Videos

Goals

  • Utilize a generative deep neural network to create deepfake videos.

  • Train a neural network architecture to accurately identify deepfake videos.

  • Input a variety of altered media content to measure the effectiveness of the proposed systems.

Takeaways

Documents