Benford’s law to detect GAN-generated images

Generative Adversarial Network (GAN) generated images are synthetic images generated by artificial intelligence. These images can be so realistic that they can fool humans into thinking that what they are seeing is real. Malicious use of this technology may seriously impact our society through fake news for example. Being able to detect whether an image is fake or real might be one of the greatest challenges against disinformation. During this talk, we will discover the strange property of Benford’s law and how it can help us discriminate against a GAN-generated image from a natural one.

Language EN
Level Level 100
Technologies

Generative Adversarial Network

Machine Learning

Lightning talk

Speaker

Tiffany Souterre
Tiffany Souterre

I love science and I love data! After finishing a Ph.D. in genetic engineering, I continued my quest for discovering new patterns through data science and machine learning. I currently work as a Data Engineer and I play with machine learning algorith...

Details