Artificial intelligence (AI) is intelligence displayed by machines. An example of this is machine learning. There are two categories of machine learning: supervised (learning by example) and unsupervised (self-learning).
Supervised machine learning requires “labels” in the data. “Labels” are the answers that enable learning by example, such as malware. Deep learning is a type of supervised machine learning which needs specialized knowledge and resources. It became famous in 2012, when a Stanford professor and Google fellow created a deep neural network to analyze millions of YouTube videos to learn to recognize images of cats.
Unsupervised machine learning doesn’t require labels in the data. In security, detecting anomalies indicative of insider threats is only possible through unsupervised machine learning. Unsupervised machine learning automatically discovers patterns from limited data sets, typically without labels.
Different machine learning techniques are effective for different types of threats. For example, to detect insider threats, unsupervised machine learning is needed.
Choosing the right technique for the problem at hand is the most important consideration for AI-enabled threat detection.
