Network security is a complex area for applying traditional machine learning. The number of possible threats is enormous, but at the same time the number of labeled attack examples is very small. Moreover, the threat vector changes when enough sample data is collected for a particular threat type.

Although it is difficult to collect examples of the truly positive results, security analysts usually have good mental heuristics about the behavior of the threats. They manually "run" the heuristics to identify the threat in the huge network data. Usually, these heuristics are applied after the unsupervised techniques identify the anomalies and outliers in the data. Although this works well in practice, the approach is computationally expensive due to the nature of the unsupervised algorithms and the unpredictable accuracy in the field.

Weak supervision offers an alternative approach to using heuristics to identify threats. It allows us to apply the heuristics to the raw data to create more efficient models with predictable accuracy. In this talk, renowned data scientist Debabrata Dash from Arista discusses a prototype for using weak supervision in cybersecurity with exciting results.

More related videos: https://www.youtube.com/playlist?list=PLZePYakcDhmgIit4s_ZP2asPdCkHSbC1X
More related videos: https://www.youtube.com/playlist?list=PLZePYakcDhmjW7BFmQe-JL5I9wykDtu64
More related videos: https://www.youtube.com/playlist?list=PLZePYakcDhmgz_Mcr2D0nS1vqhuSplq6s

#AIforcybersecurity #threatdetection #datadrivenAI

Please take the opportunity to connect with your friends and family and share this video with them if you find it useful.