Hacking Machine Learning: Towards The Comprehensive Taxonomy of Attacks Against Machine Learning Systems

Abstract

The research on machine learning systems in adversarial environments is a relatively new discipline at the intersection between machine learning and cybersecurity. Still, machine learning algorithms that beat human performance in naturally occurring scenarios are often seen as failing dramatically when an adversary is able to influence training and/or usage of machine learning system. Machine learning is already used for many extremely significant applications and will be used on a much greater scale and will have even greater significance in the approaching future. The aim of this article is to provide a comprehensive review of scientific works in the field of cybersecurity of machine learning and to present an original taxonomy of adversarial attacks against machine learning systems in this context. A pertinent taxonomy enables good understanding of full spectrum of threats and development of systems resistant to intentional hackers' attacks.