Advancement of attack and defense techniques in adversarial machinelearning

Machine Learning, these days, is helping make decisions about real-world problems and its use in almost every field signifies that it is quite effective. But, it is prone to some serious threats which are known as adversarial examples and, using these adversarial examples to attack machine learning is called adversarial machine learning. Adversaries have started to develop various methods to infiltrate machine learning models and modify them such that the models start working for their benefit or start behaving absurdly. Adversarial machine learning poses a grave threat to all the sectors in which machine learning is being used. Here, the vulnerabilities in machine learning models, major types of attacks that avail these vulnerabilities to weaken the models using adversarial examples and,defenses against these adversarial attacks are discussed.