Researchers determined that the system was able to detect 85 percent of cyberattacks, based on 3.6 million pieces of data fed to the system over a period of three months, which is a success rate about three times greater than previous benchmarks, according to Phys.org.
The algorithm works by alerting human operators of suspicious activity, who in turn decide if the threat is real. The algorithm incorporates this human input back into its prediction model so it can make better decisions the next time.
"You can think about the system as a virtual analyst," said CSAIL research scientist Kalyan Veeramachaneni, AI2 co-developer. "It continuously generates new models that it can refine in as little as a few hours, meaning it can improve its detection rates significantly and rapidly."