Towards Security Threats of Deep Learning Systems: A Survey

Towards Security Threats of Deep Learning Systems: A Survey

Overall, we summarize these attacks into four classes: model extraction attack, model inversion attack, data poisoning attack, and adversarial attack. In ...