"Adversarial Classification"
Fabrizio Ruggeri

In multiple domains such as malware detection, automated driving systems, or fraud detection, classification algorithms are susceptible to being attacked by malicious agents willing to perturb the value of instance covariates in search of certain goals.