Research Trends and Challenges on Security of Artificial Intelligence
Duan Dai, Sahar Boroomand, Xubin Xie · Zenodo (CERN European Organization for Nuclear Research) · 2020
In recent years, the development and provision of new systems and services utilizing artificial intelligence has progressed in a wide range of fields, including finance. In order to provide such a service safely, it is important to consider the security of the system (machine learning system) that implements the function of machine learning. In this paper, we show the model of machine learning system and the policy of security measures and introduce recent research examples of countermeasures against attacks in addition to the major known vulnerabilities and attack methods. Finally, we will show you the matters to keep in mind when using machine learning systems safely.