Anti-adversarial machine learning defences start to take root

Much of the anti-adversarial research has been on minute alterations to images that cause AI’s machine learning algorithms to misidentify or misclassify the images.