The Cybersecurity Dataset Observatory: Which Data for AI in Cybersecurity?
AI is increasingly used for cybersecurity tasks such as intrusion and malware detection, in enterprise, industrial, vehicular and mobile networks. Its effectiveness depends on data: models need representative network traffic, both to be trained and to be evaluated realistically. Public datasets of cybersecurity network traffic, however, are scattered across laboratory pages, repositories and aggregators, and are continuously released, corrected and withdrawn: 92 of the 234…