Applying adversarial learning to invade a spam filter

Yuanhao Luo, China 16-18

This project will apply the concepts of adversarial learning to invade a spam filter. First, techniques for invading a machine learning classifier are discussed. Next, an email spam filter based on the Naive Bayes model is proposed and realized. Then, invasion is performed on the spam filter and results are assessed. Finally, based on the invasion methods and experiment results, we can come up with security advice for the construction of a machine learning classifier.

Show more