How Neural Networks Can Be Hacked (And What You Should Do To Protect It)!

In this Tutorial I show you how easily Neural Networks can be hacked, and what you should do to protect against this. We learn about two hacking methods (Last Layer Attack and Backdooring) and I show you actual examples how to apply this. Get my Free NumPy Handbook: https://www.python-engineer.com/numpybook ✅ Write cleaner code with Sourcery, instant refactoring suggestions in VS Code & PyCharm: https://sourcery.ai/?utm_source=youtube&utm_campaign=pythonengineer * ⭐ Join Our Discord : https://discord.gg/FHMg9tKFSN 📓 ML Notebooks available on Patreon: https://www.patreon.com/patrickloeber If you enjoyed this video, please subscribe to the channel: ▶️ : https://www.youtube.com/channel/UCbXgNpp0jedKWcQiULLbDTA?sub_confirmation=1 Resources: https://github.com/Kayzaks/HackingNeuralNetworks Papers: Kang Liu et al. - Fine-pruning: Defending against backdooring attacks on deep neural networks. Shiqi Shen et al. - Auror: defending against poisoning attacks in collaborative deep learning systems. CNN Tutorial: https://youtu.be/pDdP0TFzsoQ Tensorflow Course: https://www.youtube.com/playlist?list=PLqnslRFeH2Uqfv1Vz3DqeQfy0w20ldbaV ~~~~~~~~~~~~~~~ CONNECT ~~~~~~~~~~~~~~~ 🖥️ Website: https://www.python-engineer.com 🐦 Twitter - https://twitter.com/patloeber ✉️ Newsletter - https://www.python-engineer.com/newsletter 📸 Instagram - https://www.instagram.com/patloeber 🦾 Discord: https://discord.gg/FHMg9tKFSN ▶️ Subscribe: https://www.youtube.com/channel/UCbXgNpp0jedKWcQiULLbDTA?sub_confirmation=1 ~~~~~~~~~~~~~~ SUPPORT ME ~~~~~~~~~~~~~~ 🅿 Patreon - https://www.patreon.com/patrickloeber Timeline: 00:00 - Introduction 00:42 - How To Hack Neural Nets 02:30 - Last Layer Attack 12:13 - Backdooring 16:53 - More Resources ---------------------------------------------------------------------------------------------------------- * This is an affiliate link. By clicking on it you will not have any additional costs, instead you will support me and my project. Thank you so