How I built the world’s most efficient deepfake detector with $100
Conference (BEGINNER level)
Room 4
Score 0.19
Score 0.20
Score 0.21
Score 0.21
The match becomes increasingly accurate as the similarity score approaches zero. is an ultra-realistic face generator based on neural networks, frequently used by botnets to create realistic fake profiles on the internet.

At a time when disinformation and manipulation of opinion spread through social networks, it is necessary to be equipped to spot fake identities, especially via their profile pictures which are the hardest elements to fake. During this talk, you will discover a method to detect images generated via, with 100% reliability and even a way to know the exact time the photo was produced.

The conference will focus on a technical challenge faced during the implementation of one of the attacks, which involves TensorFlow models leveraging features of a customized ElasticSearch instance, on very limited hardware.

During this beginner-friendly conference, we'll also cover the basic machine learning concepts needed to make GANs work, and a technique to generate almost undetectable fake faces yourself.

Mathis Hammel

Mathis Hammel is a tech evangelist at CodinGame, a website specialized in mini-games to learn programming. He is a specialist and technical advisor in cybersecurity, machine learning, and algorithms.

Mathis is passionate about technical challenges such as programming competitions and Capture The Flag, and holds several titles from national and international championships.

Generated Summary
WARNING: This summary was generated using GPT based on the transcript, as a result spelling mistakes and more importantly hallucinations can be present.

Deep Fakes and Artificial Intelligence
Generating Random Faces with AI
The speaker discussed a website called \"This Person Does Not Exist\" which uses Artificial Intelligence (AI) to generate random faces. A detector has been built to detect whether or not these faces are real. To demonstrate this, the speaker played a game with the audience, showing that AI can easily fool humans.
The Implications of Deep Fakes
The speaker also explained the implications of deep fakes, such as their use for political manipulation and misinformation, and how they can be used to create fake social media profiles.
Combatting Deep Fakes
The speaker discussed how to combat deep fakes by hacking the website used to generate them and detecting whether or not a profile is legitimate. This includes using facial recognition to detect slight changes to the image, such as rotation, flipping, or noise, and determining when an image was generated.
Neural Networks and GANs
This is a brief crash course on how Artificial Intelligence works using neural networks. Neural networks are a sequence of multiplications and additions which can be trained to create specific results. Generative Adversarial Networks (GANs) are a type of neural network used to generate things such as faces, and they are trained by having two networks work together: a generator which creates outputs from random numbers, and a discriminator which determines whether the image is real or not.
Testing This Technique
This summary explains how to detect if a social media profile is fake or not by looking at the position of the eyes in the profile picture. It suggests testing this technique on Twitter and Facebook to check for suspicious profiles.
This talk provided an overview of how Artificial Intelligence works, specifically in terms of deep fakes and neural networks. It also discussed the implications of deep fakes and how to combat them. Finally, it provided a technique for detecting fake social media profiles by looking at the position of the eyes in the profile picture. In conclusion, it is important to be aware of the implications of deep fakes and to be vigilant about detecting them.
You can also ask questions on the complete talk using Devoxx Insights