adversarial.js: Break any neural network live in your browser - Made with TensorFlow.js

Today on Made With TensorFlow.js we’re joined by Kenny Song, an active researcher on security and reliability, where he shows you how to break neural networks in your web browser in real-time by changing inputs, such as pixels in an image, to fool a machine learning model. Watch as he turns a photo of a “dog” which is initially classified correctly to be misclassified as a “hotdog” - even though to you, as a human, the image still looks the same. Learn how he does it in this educational video so you can make your models even more robust to such attacks in the future. Try the adversarial.js demo yourself → https://goo.gle/3cHLcCP​ Get the source for adversarial.js on GitHub → https://goo.gle/39DZNND​ Want to be on the show? Use #MadeWithTFJS to share your own creations on social media and we may feature you in our next show! Catch more #MadeWithTFJS interviews → http://goo.gle/made-with-tfjs Subscribe to the TensorFlow channel → https://goo.gle/TensorFlow