With the increasingly severe form of cyberspace confrontation, cyberspace security has been paid more and more attention by all countries in the world and presents the development trend from feature-based “identity recognition” to identity-based “identity recognition.” Through the research of cyberspace security development, machine learning, and cyberspace security confrontation drill method, the feasibility of dual role multilevel confrontation method is demonstrated. In this paper, a preliminary scheme of applying machine learning to cyberspace security is proposed, and the concrete implementation of dual role multilevel confrontation method for cyberspace security and the specific application of machine learning in cyberspace security are given. It improves the security and credibility of cyberspace itself and improves the ability of cyberspace to resist attacks.