Labelled Data-as-a-Service

Research in cyber-security is often hampered by a lack of access to ground-truth data of malware behavior in the wild. Primarily, malware is as dangerous and stealthy as a ticking time bomb, adding significant challenges to collecting their real-world behavior without being affected by its consequences. We present RaDaR, an open real-world dataset for malware behavioral analysis, with mechanisms to keep pace with the evolving malware landscape. With a comprehensive view of malware activity across the system and diverse perspectives for analyzing them, RaDaR has multiple use cases for AI-based security research, including developing novel multi-featured countermeasures and unbiased comparison of detection approaches. By abstracting the challenges in malware data collection, RaDaR opens up the cyber-security field by enabling not only security researchers, but other communities, especially data science researchers, to explore and analyze it quickly.