Hacking neural networks using adversarial examples
Bogdan Gliga - Associate Software Engineer @ Telenav
Europa room, 3rd floor
12nd November, 15:40-16:10
In this presentation we will talk about how we can trick neural networks into behaving how we want. We will see both how we can create specifically designed images which wil act as "optical illusions" for the neural nets and how to create algorithms which are robust to these kind of attacks
I am very passionate about everything related to AI & Machine Learning. As I trully believe that these technologies will massively impact and change the society that we live in, I am always eager to learn about new AI frameworks, investigate interesting datasets and further my knowledge in the field. Currently, I work at Telenav, being one of the engineers in the Computer Vision team, where I help develop our AI platform in order to extract useful features from images in order to improve our maps.