Computer Security and Machine Learning: Worst Enemies or Best Friends?

摘要: Computer systems linked to the Internet are confronted with a plethora of security threats, ranging from classic computer worms involved drive-by downloads and bot networks. In last years these threats have reached new quality automatization sophistication, rendering most defenses ineffective. Conventional measures that rely on manual analysis incidents attack development inherently fail provide timely protection threats. As consequence, often remain unprotected over longer periods time. The field machine learning has been considered an ideal match for this problem, as methods ability automatically analyze data support early detection However, only few research produced practical results so far there is notable skepticism in community about learning-based defenses. paper, we reconsider problems, challenges advantages combining security. We identify factors critical efficacy acceptance present directions perspectives successfully linking both fields aim at fostering intelligent methods.