Machine learning

ESET has been experimenting with machine-learning algorithms to detect and block threats since 1990s, with neural networks making their way into our products already in 1998. Since then we have implemented this promising technology all across our multi-layered technology.

This includes DNA detections, which use models based on machine learning to work effectively with or without cloud connection. Machine-learning algorithms are also a vital part of the initial sorting and classification of incoming samples as well as of placing them on the imaginary “cyber-security map”. But most importantly, ESET has developed its own in-house machine-learning engine dubbed ESET Augur. It uses the combined power of neural networks (such as deep learning and long short-term memory) and a handpicked group of six classification algorithms. This allows it to generate a consolidated output and help correctly label the incoming sample as clean, potentially unwanted or malicious.

ESET Augur engine is fine-tuned to cooperate with other protective technologies such as DNA, sandbox and memory analysis as well as with the extraction of behavioral features, to offer the best detection rates and lowest possible number of false positives.