Manipulating hidden-Markov-model inferences by corrupting batch data

William N. Caballero, José Manuel Camacho, Tahir Ekin, Roi Naveiro

Abstract: Highlights•The adversarial machine learning perspective is extended to hidden Markov models.•Three corruption problems are developed to attack inference on hidden Markov models.•Multiple novel attacks are developed for application in a grey-box setting.•Extensive computational testing verifies attack efficacy and compares performance.