Security is overdue for actionable forecasts. Like predicting the weather, similar models should work for vulnerabilities. With some open source data and a clever machine learning model, Kenna Securities can predict which vulnerabilities attackers are likely to write exploits for. Their model has 90 percent accuracy, one the day a vulnerability is released. The speaker will issue some forecasts live.
Learning Objectives: 1: Learn about predictive models for vulnerability data. 2: How to evaluate a machine learning model in our context—vulns, exploits. 3: How to select the proper machine learning model, how to ask the right question.